European Journal of Pharmaceutics and Biopharmaceutics 71 (2009) 2337

Contents lists available at ScienceDirect

European Journal of Pharmaceutics and Biopharmaceutics

journal homepage: www.elsevier.com/locate/ejpb

Review article

Solid form screening A review

Jaakko Aaltonen a,*, Morten Alles b, Sabiruddin Mirza c, Vishal Koradia b, Keith C. Gordon d, Jukka Rantanen b
a

School of Pharmacy, University of Otago, Dunedin, New Zealand

Department of Pharmaceutics and Analytical Chemistry, University of Copenhagen, Denmark
Division of Pharmaceutical Technology, University of Helsinki, Finland
d
Department of Chemistry, University of Otago, Dunedin, New Zealand
b
c

a r t i c l e

i n f o

Article history:
Received 29 February 2008
Accepted in revised form 18 July 2008
Available online 31 July 2008
Keywords:
Amorphous
Hydrate
Polymorph
Polymorphism
Screening
Solid form
Solid state
Solvate

a b s t r a c t
Solid form screening, the activity of generating and analysing different solid forms of an active pharmaceutical ingredient (API), has become an essential part of drug development. The multi-step screening
process needs to be designed, performed and evaluated carefully, since the decisions made based on
the screening may have consequences on the whole lifecycle of a pharmaceutical product. The selection
of the form for development is made after solid form screening. The selection criteria include not only
pharmaceutically relevant properties, such as therapeutic efcacy and processing characteristics, but also
intellectual property (IP) issues. In this paper, basic principles of solid form screening are reviewed,
including the methods used in experimental screening (generation, characterisation and analysis of solid
forms, data mining tools, and high-throughput screening technologies) as well as basics of computational
methods. Differences between solid form screening strategies of branded and generic pharmaceutical
manufacturers are also discussed.
2008 Elsevier B.V. All rights reserved.

1. Introduction and background

In a classic paper by Haleblian and McCrone, polymorphism was
dened as the ability of any compound to crystallise as more than
one distinct crystal species [1]. However, in physical pharmacy the
word polymorphism is nowadays often used to cover a variety of
solid forms of active pharmaceutical ingredients (APIs) and excipients including crystalline, amorphous, and also solvate/hydrate
forms. Accordingly, the activity of generating, isolating and analysing different solid forms of an API is known as polymorph screening. In this paper, to avoid term confusion, all above-mentioned
solid modications are referred to as solid forms. The aim is to give
a pharmaceutical outlook on solid form screening what it consists of and what is the signicance of it for the pharmaceutical
eld. As such, focus is not put on the screening and selection of
salts and co-crystals.
1.1. Solid form screening
Nowadays solid form screening is a standard procedure in drug
development, but it was not until the last decades of the 20th century that the whole pharmaceutical industry became aware of
* Corresponding author. School of Pharmacy, University of Otago, P.O. Box 913,
Dunedin 9054, New Zealand. Tel.: +64 3 479 7272; fax: +64 3 479 7034.
E-mail address: [email protected] (J. Aaltonen).
0939-6411/$ - see front matter 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.ejpb.2008.07.014

polymorphism even though the phenomenon had been known

since the early 19th century [2]. The patent cases [3] and enormously expensive HatchWaxman agreements between branded
and generic drug manufacturers [4], and production problems such
as the sudden appearance of another form of ritonavir (with low
solubility and bioavailability) [5] were needed for the industry to
start taking polymorphism seriously. Regardless of the fact that
solid form screening is a regulatory requirement for new pharmaceuticals [6], the whole topic remains still somewhat open since no
universal guidelines on solid form screening can be written
because every compound possesses unique properties. Further,
all polymorphism-related phenomena, e.g., nucleation, are not
yet fully understood [7,8].
The aim of solid form screening is to nd the optimal form with
the best characteristics for development. In order to be able to select the optimal form, knowledge of as many as possible forms is
needed. The choice of the form for development is usually a compromise between physical, chemical, pharmaceutical and biopharmaceutical properties (see Section 1.2). Traditionally, the most
stable form is favoured over other forms because of its lower tendency to solid phase transformations. It is important to identify the
stable form as early as possible in the drug development process to
avoid subsequent setbacks [9,10]. However, metastable forms are
sometimes deliberately chosen usually for better solubility and
thus bioavailability [11]. Solid form screening should not be a
one-time effort performed only during the preformulation stage

24

J. Aaltonen et al. / European Journal of Pharmaceutics and Biopharmaceutics 71 (2009) 2337

of drug development. Instead, the solid state of the API should be

monitored in a continuous fashion also during scale-up and manufacturing, to detect the occasional event of late appearing polymorphs as early as possible.
Solid form screening can be approached experimentally and
computationally. Experimental screening consists of the preparation step, during which various forms are generated and isolated
(see Section 2.1), and the analysis step, which involves the use of
various measurement techniques (see Section 2.2) and also the
data analysis (see Section 2.3). Computational methods of polymorph prediction (see Section 3) have evolved markedly during
the last years, but one cannot fully rely on them yet. Even if the
crystal structures are not predictable, computational methods
can be used to help rationalise the experimental procedures [12]
and decide whether the stable form has been found or not [13]. Solid form screening is one of the cornerstones of drug development
and plays an important part in pharmaceutical product lifecycle
management [14]. If most of the relevant forms are found the innovator company can achieve a very valuable intellectual property
(IP) situation. Poorly conducted screens and unsuccessful patenting strategies on the other hand open new possibilities for competitors (see Section 4).
1.2. Differences between solid forms and their implications on physical,
chemical, pharmaceutical and biopharmaceutical properties
In the crystalline state (polymorphs, solvates/hydrates, co-crystals), the constituent molecules are arranged into a xed repeating
array built of unit cells, which is known as lattice, whereas in the
amorphous state there is no denite long-range order [15,16].
Polymorphs contain molecules of only one chemical species in
their unit cells. There are two ways in which crystal lattices can
form: packing polymorphism and conformational polymorphism.
If rigid molecules with a specic conformation are packed in different arrangements, then it is referred to as packing polymorphism,
whereas conformational polymorphism denotes crystal forms that
consist of exible molecules with different conformations packed
in different arrangements [16]. See paracetamol [17] and L-glutamic acid [18] for examples of packing and conformational polymorphism, respectively.
In case the unit cell is built of host molecules (the API) accompanied by guest molecules the resulting solid form is called a solvate or a co-crystal depending on whether the guest species is
liquid or solid at ambient conditions, respectively [19]. If the guest
molecule is water then the term hydrate is used. (There has been,
and still is, controversy in the literature regarding the nomenclature of multi-component crystals.) Solvates and co-crystals can
also exhibit polymorphism. Hydrates are the most common solvates and deserve special attention. Many APIs are capable of forming hydrates due to the small size and multidirectional hydrogen
bonding capability of the water molecule [20,21]. Hydrate formation stabilises the crystal structure via intermolecular bonding in
the crystal lattice, resulting in hydrates being the stable forms in
aqueous surroundings (below dehydration temperature). Many hydrates are also stable in ambient conditions, and therefore are
quite often chosen as the form for development. The hydrate form
can be chosen for development to avoid hydrate formation during
downstream processing, but then again with hydrates there is a
risk of dehydration. A classication system of hydrates has been
suggested [22].
Amorphous state can be the result of either physical manipulation or the intrinsic nature of the compound [23]. The amorphous
form is the most soluble form, but on the other hand, it is also the
form with the lowest stability. The current trend of new APIs getting less and less soluble has given rise to the use of metastable
forms in formulations and also another research topic, stabilisation

Table 1
Physical properties that differ among various solid forms
Packing properties
Thermodynamic
properties
Spectroscopic
properties
Kinetic properties
Surface properties
Mechanical
properties

Unit cell volume (crystalline forms only), density, refractive

index, hygroscopicity
Melting point, enthalpy, entropy, free energy, solubility
Electronic transitions (UVvis spectra), vibrational transitions
(IR and Raman spectra), rotational transitions (far-IR or
microwave spectra), nuclear spin transitions (NMR spectra)
Dissolution rate, rates of solid-state reactions, stability
Surface free energy, interfacial tensions, crystal habit
Hardness, tensile strength, compactibility, tableting,
owability

Modied from [16].

of metastable forms and formulations thereof [18,24,25]. The physical stability issues are the main hurdles in developing formulations of amorphous APIs [26]. Regardless of being stable in dry
conditions over the whole shelf-life, metastable forms may rapidly
convert to the stable form upon administration, since solvent-mediated solid phase transformation kinetics are faster than those of
solid-solid transformations. Therefore, the solubility advantage of
amorphous (and other metastable) forms may not always be fully
exploited [27].
The solid forms of a given API can have signicantly different
physicochemical properties that can affect its performance [1,28].
Some of these properties are listed in Table 1. If solubility and/or
dissolution rate are dependent on the solid form, the bioavailability of the API can be affected. This is a particularly important note
when developing BCS class II APIs (low solubility and high permeability) with dissolution dependent bioavailability [29]. Examples
of APIs with bioavailability problems due to solid-state phenomena
are carbamazepine [30,31] and ritonavir [5]. Mechanical property
differences can affect processing behaviour, and this is the case,
for example, with paracetamol [32]: direct compression of form
II is feasible, whereas with form I, binder excipients have to be
used [33]. Also different forms of theophylline [34] and sulfamerazine [35] have been reported to show different processing characteristics. Stability is a very important property of a solid form,
considering that raw materials and pharmaceutical products may
be stored for prolonged periods and the solid state must remain
unchanged. In addition to physical stability, chemical stability also
has to be taken into account. Chemical reactivity can vary between
different solid forms, and sometimes certain solid forms can be
used to cause reactions when desired or to prevent reactions when
they are to be avoided [36]. For examples of differences in chemical
stability between forms see prednisolone tert-butylacetate [37]
and quinapril HCl [38].
2. Experimental solid form screening
2.1. Generation of solid forms
2.1.1. Basics thermodynamics and kinetic effects
Crystallisation is the key experimental technique used to execute solid form screens. Within this review, crystallisation is considered as a tool to generate multiple solid forms of a
pharmaceutical compound; the fundamental theoretical aspects
of this process can be found elsewhere [39,40]. Classically, the
crystallisation process is described in terms of two distinct steps,
nucleation and crystal growth [41], with the resulting physical
form being the consequence of the kinetic relationship between
these two elementary processes. In other words, for a polymorphic
system the polymorph that nucleates rst is thought to come from
the cluster that exhibits the fastest nucleation rate as a result of its
lowest free energy barrier (DG) to nucleation. However, the nature

J. Aaltonen et al. / European Journal of Pharmaceutics and Biopharmaceutics 71 (2009) 2337

of the polymorph that eventually crystallises is determined by the

combination of the relative nucleation rates and the relative crystal
growth rates of the polymorphs [42,43]. Hence, nuclei of different
structures can form and coexist in a given crystallisation leading to
a mixture of crystalline forms in the resulting solid when kinetic
factors prevent the achievement of equilibrium. It is therefore a
general rule that the stable form is likely to be obtained when
operating under thermodynamic conditions (e.g., slow cooling),
while a metastable form is expected to be produced under kinetic
conditions (e.g., rapid cooling). In the context of polymorph screening, this implies that the interplay between kinetic and thermodynamic factors has to be extensively employed to discover all the
relevant solid forms of a compound in question. The possible scenarios of crystallisation in a dimorphic system are schematically
presented in Fig. 1.
2.1.2. Different ways to generate solid material
There are a large number of classical techniques that can be
used to generate solid materials [25,40,44,45] (Table 2). Since
one approach can favour the nucleation and growth of one form

25

Table 2
Methods to generate various solid forms [25,40,44,45]
Method

Degrees of freedom

Crystallisation by
cooling a solution
Solvent evaporation

Solvent, cooling prole, concentration, mixing

Precipitation
Vapour diffusion
Suspension
equilibration
Crystallisation from
the melt
Quench cooling the
melt
Heat induced
transformations
Sublimation
Desolvation of solvates
pH change
Mechanical treatment
(i.e., milling,
cryo-grinding)
Freeze-drying
Spray drying

Solvent, initial concentration, evaporation rate,

temperature, pressure, ambient relative humidity
Solvent, anti-solvent, rate of anti-solvent addition,
mixing, temperature
Solvents, temperature, concentration
Solvent, temperature, solubility, temperature programs,
mixing, equilibration time
Temperature changes (min, max, gradients)
Cooling rate
Temperature changes
Temperature gradient, pressure, surface type
Temperature, pressure
Temperature, rate of change, acid/conjugate base ratio
Milling time, mill type
Solvent, concentration, temperature programs
Solvent, concentration, drying temperature

over another, it is essential to perform screening experiments by

a variety of methods under various process conditions. In general,
crystallisation from solution (cooling or evaporative, as well as
slurry conversion) and recrystallisation from a neat compound
(sublimation, thermal treatment, crystallisation from the melt,
grinding, and thermal desolvation) are the methods of choice for
solid form screens [45,46]. Crystallisation from solution is customarily used in solid form screens for several reasons. Firstly, a large
number of polymorphs and solvates can be discovered by changing
the solvent system. Secondly, pharmaceutical solids are often exposed to different solvents during processing, and thus the solvate/hydrate formation tendency of APIs should be systematically
studied in order to design the manufacturing processes. Thirdly,
desolvation of solvates or dehydration of hydrates is another useful, and sometimes the only, technique to discover a polymorphic
form [47,48]. Finally, a solvate (and especially a hydrate) can be
of interest as a commercial product.
Recently, several innovative techniques for example, capillary
crystallisation [49], laser-induced crystallisation [50], and sonocrystallisation [51] that promote nucleation and hence discovery
of alternate crystalline forms have been reported. Another recent
solid-state topic of growing interest is the design and formation
of co-crystals. Co-crystallisation opens new ways to produce solid
forms with unique characteristics for, e.g., combination therapy
and IP protection. For deeper insight into the preparation of cocrystals the reader is referred to texts by Rodrguez-Hornedo
et al. [52] and Stahly [53].

Fig. 1. Possible scenarios of crystallisation in a dimorphic system under different

conditions. Polymorph A is the stable form with a higher free energy barrier (DG) to
nucleation.

2.1.3. Critical parameters in crystallisation of polymorphs

As crystallisation is a highly complex process there are several
process variables that can affect the outcome (Table 3). Supersaturation as the driving force of crystallisation is the key thermodynamic
variable that affects the kinetics of crystal nucleation and growth. In
other words, the resulting crystalline form may vary with the degree
of supersaturation. Temperature can be considered as the second
most signicant variable affecting crystallisation outcome in a polymorphic system. The effect of temperature has both thermodynamic
and kinetic implications, particularly for enantiotropic polymorphs
which change the solubility order near the transition temperature
[54]. Practically speaking, crystallisation at one temperature may
produce one polymorph, while crystallisation at another temperature may yield a second polymorph [55]. For some solvate systems,

26

J. Aaltonen et al. / European Journal of Pharmaceutics and Biopharmaceutics 71 (2009) 2337

Table 3
Process variables affecting the outcome of crystallisation from solution [46]
Crystallising phases

Crystallisation method

Polymorphs/solvates

Salts/co-crystals

Cooling crystallisation

Evaporation

Precipitation

Slurry conversion

Degree of supersaturation
Solvent composition
Additive type

Counter-ion type
Acid/base ratio
Solvent/Solvent combination

Heating rate
Cooling rate
Maximum temperature

Evaporation rate
Evaporation time
Carrier gas

Solvent type
Incubation temperature
Incubation time

Additive concentration

Degree of supersaturation

Incubation time

Surface-volume ratio

Anti-solvent type
Rate of anti-solvent addition
Temperature of
anti-solvent addition
Time of anti-solvent addition

Additive type
Additive concentration
pH
Ionic strength

Incubation temperature

it can be anticipated that by changing the harvesting temperature

the solvates of various stoichiometry will be obtained, with lower
temperatures favouring the formation of solvates of higher stoichiometry [20]. Therefore, crystallisations under various temperature
proles should be performed during solid form screens. Solvent,
additives (and impurities), interface, pH and host-guest composition
have been classied as secondary factors affecting the crystallisation
outcome [56], mainly through their effect on the degree of supersaturation (Fig. 2).
Selecting the right set of solvents is important when designing
solid form screening experiments. Selection criteria should encompass (1) solvents with broad distribution of properties [5759], (2)
potential solvents used in synthesis, purication and processing,
(3) solvents and excipients used in the nal formulation. (E.g., if
the pharmaceutical product is formulated in a soft gelatin capsule,
the carrier medium should be included in the screening.) In addition, crystallisations from water and water-solvent mixtures are
usually included in the screens to generate hydrates. Papers dealing with the analysis of multivariate solvent databases have been
published to aid the selection of solvents [57,59,60].
In summary, to increase the probability of discovering all the
relevant forms, the multiparameter space that contributes to solid
form diversity should be covered as broadly as possible. This is
usually achieved by designing a rational set of the process variables [61].
2.1.4. Seeded and additive-mediated crystallisations
Seeding (the addition of solid particles of the desired phase to a
crystallisation medium) is a common approach to crystallise the

Solubility of polymorphs

Control of crystallisation
of polymorphs

Nucleation
Crystal growth
Phase transformations

Primary factors:

Secondary factors:

Supersaturation
Temperature
Agitation
Seed crystals

Solvent composition
Additives/Impurities
Interface
pH
Host-guest composition

Fig. 2. Hierarchy of the process parameters controlling crystallisation outcome in

polymorphic systems. Modied from Kitamura et al. [56].

Thermal cycling and

gradients

desired crystalline form [62]. Controlling crystallisation outcomes

via seeding relies on the potential of crystal surfaces to promote
heterogeneous or secondary nucleation, while avoiding heterogeneous nucleation mediated by unknown contaminants [40]. In
the context of solid form screening, this implies that some precaution measures (e.g., ltration of a supersaturated solution) should
be undertaken to avoid the crystallisation mediated by the initial
solid phase. Conversely, it may be useful to employ heterogeneous
nucleation [18,24,63,64] mediated by various additives (e.g., structurally related compounds or polymers) as a solid form screening
tool. In such cases, the additive acts as an additional diversity element which affects the crystallisation outcome, and thus provides
the potential for the discovery of unknown solid forms without
prior knowledge of the crystal structure. For example, a fourth
polymorph of carbamazepine, which appeared to be more stable
than the well-studied trigonal form [65], was discovered using
the polymer heteronucleation approach [66].
2.1.5. High-throughput screening technologies
High-throughput (HT) screening utilises fully automated robotic systems capable of performing thousands of crystallisations
per week with only few grams of API consumption [44,67]. It is believed that the immense number of crystallisation trials will increase the probability of nding solid forms suitable for further
development and/or patenting. HT screening is most commonly
carried out in 96-well plate systems in which the particular solid,
dissolved in a suitable solvent, is initially dispensed using automated liquid handling systems. The amount of API is reported to
lie in the range of 0.510 mg per sample [6769]. Different levels
of supersaturation can be achieved by, for example, heating/cooling, evaporation, and by varying the nominal concentration of
API. Similarly to bench-scale crystallisations, slurry experiments
can also be implemented in HT systems. The resulting solid is commonly analysed in-situ using either XRPD or Raman spectroscopy.
The generated forms are thereafter identied using one or several
suitable data mining tools (see Section 2.3) and the results are
stored in a database for later use. It should be noted that microscale in-situ analysis is relatively prone to sampling errors, and
may thus lead to bad quality data or false results. Very often follow-up studies (e.g., including up-scaled API quantities) are performed for the purpose of further elucidating interesting solid
forms found in the initial HT screening or in an attempt to nd
new forms by further varying the crystallisation conditions. As an
example of this, Peterson et al. demonstrated an approach to the
so-called iterative HT screening comprising a multitude of experiments under varying as well as similar conditions [70]. Acetaminophen (paracetamal) was subjected to screening and three
experimental runs/iterations of evaporative crystallisations were
required to identify and further test the reproducibility of Form II
formation (starting material was Form I). It was only possible to
generate Form III by subsequent melt crystallisation. Conditions

27

J. Aaltonen et al. / European Journal of Pharmaceutics and Biopharmaceutics 71 (2009) 2337

and results for a series of HT polymorphism studies are summarised in Table 4. It is important to emphasise that in spite of the
vast number of crystallisations, the hit rate (i.e., the percentage
of vials containing API precipitates) is often very low. Transform
Pharmaceuticals have reported hit rates between 2.5% and 13%
for HT screening of various APIs [6971]. Clearly, the ultimate
advantage of HT solid form screening lies in the large number of
experiments carried out, which in addition utilise low API quantities and little if no manual intervention. Automated approaches do,
however, suffer from some drawbacks as opposed to manual
bench-scale crystallisations. Firstly, there is limited ability to incorporate some of the pivotal crystallisation methods other than the
basic solvent-based techniques (see Table 2). Secondly, a large fraction of the found solid forms are quite often not true polymorphs
but rather solvates, and therefore less useful for further development (although this information may still be benecial from a
manufacturing point of view) [72]. Lastly, subsequent up-scaling
of solvent volumes, as a part of further stability studies or API manufacturing, can change the polymorphic outcome markedly.

It is generally impossible to guarantee that all solid forms of a

compound in question will be discovered by HT screening since
there is no method that provides complete exploration of the solid-state landscape. It is therefore recommended to supplement
the results of the HT studies with more in-depth crystallisation
experiments.
2.2. Identication and analysis of solid forms
There is a wide array of methods available for solid phase analysis of pharmaceutical materials, and some excellent books and
book chapters on the methods have been published [7375]. Techniques commonly used to study solid-state properties are listed in
Table 5 [76]. The method of choice for a specic case depends on
the key parameters one needs to determine and how deeply they
have to be investigated. Usually it is advisable to use two or more
complementary methods to obtain a reliable knowledge of the
forms, but as mentioned in the previous chapter the amount of
samples produced by HT screening methods can be very large

Table 4
Conditions and results for a series of HT screening studies
Reference

API

Platform/company

No. of crystallisations/
no. of solvents

HT crystallisation
method

HT analytical
method

No. of forms found

Park et al. [67]

SSCI, Inc.

XRPD, Raman

Cimetidine

Cooling, evaporation,
cooling and evaporation
Cooling, evaporation,
precipitation

XRPD

Desrosiers [68]

Chemspeed Accelerator
SLT100
Symyx technologies, Inc.

3 polymorphs 3 solvates
Amorphous phase
3 polymorphs (Forms I, II
and III) 9 solvates
3 polymorphs

Hilker et al. [89]

Carbamazepine

Solvias AG

288 crystallisations/
solvents N/A
594 crystallisations/66
solvents
288 crystallisations/20
solvents, 84 solvent
mixtures
Crystallisations N/A/43
solvents and mixtures

N/A

Florence et al. [72]

Buspirone
hydrochloride
Carbamazepine

Almarsson et al. [71]

CrystalMax /TransForm
Pharmaceuticals, Inc.

1440 crystallisations/
21 solvents and
mixtures

Cooling (three nominal

concentration levels),
separate hydrate screen

Raman, XRPD

Almarsson et al. [71]

Angiotensin II
receptor
antagonist
MK-996
Sertraline HCl

CrystalMax /TransForm
Pharmaceuticals, Inc.

Cooling (two nominal

concentration levels)

Raman, XRPD

10 forms (4 additional
forms by follow up studies)

Morissette et al. [69]

Ritonavir

CrystalMax /TransForm
Pharmaceuticals, Inc.

3072 crystallisations/
24 solvents and
mixtures
2000+ crystallisations/
24 solvents (single and
binary mixtures)

Cooling (four nominal

concentration levels)

Raman

Peterson et al. [70]

Acetaminophen

CrystalMax /TransForm
Pharmaceuticals, Inc.

7776 crystallisations
(rst iteration)/16
solvents (single and
binary mixtures)

Cooling (three nominal

concentration levels)

Raman

3 polymorphs (I, II, IV) 1

solvate (III). Follow up
studies revealed a
trihydrate (Form V)
2 polymorphs (I and II).
Form III was found in
subsequent studies

TM

TM

TM

TM

Raman, XRPD,
polarising light
microscopy
Evaporation, suspension, Raman
desolvation

4 known polymorphs 2
known solvates (dihydrate
and acetone) 5 suggested
new forms (incl. 2 solvates)
18 crystalline forms

Table 5
Methods to study solid-state properties
Method

Data measured

Property measured/use

X-ray diffraction (single crystal and XRPD)

Infrared (IR) spectroscopy
Raman spectroscopy
Terahertz pulsed spectroscopy (TPS)
Near-infrared spectroscopy (NIR)
Solid-state NMR
Differential scanning calorimetry (DSC)
Thermogravimetry (TG)
Microscopy, PLM, SEM

Diffractogram
IR spectrum
Raman spectrum
Terahertz pulsed spectrum
Near-infrared spectrum
Magnetic resonance
Heat ow vs. temperature
Change of mass vs. temperature
Microscopy under the inuence
of light or electron radiation
Change of mass vs. variable RH%

Crystallographic properties
Chemical information
Chemical information (complementary to IR)
Chemical information, lattice phonon modes
Chemical information (overtones and combinations of IR vibrations)
Chemical information
Thermal events
Solvate/hydrate studies
Morphology, surface examination, dehydration, polymorphism (PLM)

Moisture sorption/desorption isotherms

Solubility/dissolution
Microcalorimetry
Solution calorimetry

Amount dissolved in different solvents

or temperatures/vs. time
Heat ow vs. time
Heat ow during dissolution

Hygroscopicity behaviour (hydrate formation,

dehydration, amorphous crystallisation)
Solubility/dissolution rate measurement
Quantication of amorphous form
Quantication of polymorphs and amorphous form

28

J. Aaltonen et al. / European Journal of Pharmaceutics and Biopharmaceutics 71 (2009) 2337

and the use of multiple methods of analysis may not be reasonable

timescale wise. Instead of using different methods to complement
each other, it may be wise to use hyphenated techniques that permit complementary data to be acquired in a single measurement.
In addition to the identication and quantication of the solid
forms, the hygroscopicity and moisture sorption behaviour as well
as solubility and dissolution rate of the solid forms need to be
tested to be able to select the optimal form. Moisture sorption analysers [77,78] and dissolution apparatuses [79] can be interfaced
with in-situ spectroscopy to achieve better understanding of the
solid-state phenomena during the analysis. The intrinsic dissolution rate (IDR) test [80] is a good way to investigate the effects of
formulation excipients and gastrointestinal uids (by the use of
biorelevant dissolution media) on various solid forms and possible
solid-state transformations occurring during dissolution. Generally
speaking, it is wise to investigate the effect of excipients on the solid-state stability in both dry and wet conditions.
2.3. Data mining in solid form screening
An inevitable consequence of solid form screening is the generation of either small or large amounts of data. Effectively extracting
relevant information from this data pool is no easy task for the researcher, considering the myriad of existing data analysis tools,
ranging from simple visual inspection to unsupervised or supervised multivariate analysis. It is not possible to identify one gold
standard for data analysis in solid form screening; instead the specic choice of approach depends on (1) the features of the data matrix with respect to matrix size and presence/absence of
independent variables, (2) instrumentation used for data acquisition and (3) user preferences. In general, data treatment may be
divided into the following categories: design of experiments, visual
inspection of data, unsupervised and supervised analysis (Fig. 3). It
is important that the researcher understands the possibilities and
limitations of each category before using the data analytical tools.
2.3.1. Design of experiments
Before initiating any experiment, careful considerations should
be paid as to how it should be carried out, that is, the construction
of a statistical design. Failure to do so may result in poor quality
data with little, if no, possibility of performing efcient and meaningful data analysis. The goal of design of experiments (DoE) is to
plan and conduct experiments that allow extraction of a maximum
amount of information from the collected data in as few runs as

possible. As part of the DoE, (expected) critical process parameters/factors and their settings/levels are identied [81]. Ultimately,
this will permit assessment of how each factor and their interactions affect a certain response variable through suitable statistical
analysis. As reviewed by Yu et al. [82], DoE may be particularly
useful for monitoring and controlling complex crystallisation processes, involving several critical process variables such as heating/
cooling proles, solvent type, solvent evaporation rate proles, and
seeding. These factors are highly interrelated and will affect the solid-state outcome of the crystallisation process. Thus, DoE can help
unravel the complex nature of these and clarify their impact on the
nucleation and crystallisation of several crystal forms at the early
stage of solid form screening. An identical setup (merely implemented at larger scale) may be applied at a later stage for the controlled manufacturing of a desired API crystal form [82]. Al-Zoubi
et al. [83] implemented a full factorial design to elucidate the effects of cooling temperature and harvesting time, each at three levels, on the formation and quality of orthorhombic paracetamol
(Form II polymorph) via solution-mediated transformation. Data
analysis on the factors and their interactions was carried out using
ANOVA and revealed signicant effects of both crystallisation
parameters on several response variables as well as signicant
interaction. In a related study, the effects of various parameters
on the solution-mediated polymorphic transformation of buspirone hydrochloride were investigated [84]. This experiment utilised a 24 factorial design having pH, solvent composition,
amount of co-solvent, and impurities (seeding) as factors and degree of interconversion of metastable form 2 to stable form 1 as
a response variable. The data were subjected to response surface
modelling to conveniently visualise the effects of the four factors.
From here, pH and amount of co-solvent and their interaction were
found to have the strongest effect on the interconversion to form 1.
Higher-order interactions are very common in chemical processes
[85]. Unfortunately, interpretation of more than two-way interactions is almost impossible using traditional multi-way ANOVA and
more sophisticated modelling tools are needed. In this context,
GEMANOVA (GEneralised Multiplicative ANOVA) is suggested as
a possible alternative to ANOVA for analysing complex data effectively and for obtaining more interpretable solutions enabling the
overview of the whole sampling region. So far, GEMANOVA has
been used mainly within the food industry [86]. However, it is expected to provide distinct advantages for the elucidation of complex pharmaceutical processes, such as crystallisation, in which
the researcher wishes to understand and visualise the effect of

Fig. 3. Data analysis suggestions for solid form screening.

J. Aaltonen et al. / European Journal of Pharmaceutics and Biopharmaceutics 71 (2009) 2337

interacting process variables on, for instance, polymorphic

outcome.
2.3.2. Visual inspection of data
Visual evaluation of experimental raw or pre-processed data is
an often utilised approach in many areas of research and not least
in the investigation of solid-state phenomena. Evidently, manual
assessment seems most rational in cases where it is practically feasible, that is, when the amount of data is limited and/or if well-resolved characteristic peaks are present. It is important to point out
that visual inspection of spectral data can sometimes be tedious,
because the differences between the spectra of various solid forms
can be very small. In cases where manual examination of data is
unfeasible, it might be more rational to compare entire spectra
by simple pair-wise subtraction, thereby permitting a more convenient overview of the spectral regions of interest for differentiating
between the solid forms.
Visual inspection remains a powerful solution to the task of
data interpretation. Even when applying sophisticated data modelling tools (see below), it is still highly recommended to support the
ndings by visual inspection of the relevant raw data. This greatly
reduces the risk of drawing erroneous conclusions from the solid
form screening experiments.
2.3.3. Unsupervised methods
Unsupervised analysis of data is performed when no a priori
information on the solid material under investigation is available.
This is often the situation during early screening of API candidates,
where little is known about the solid-state properties [87]. In the
current review, unsupervised methods are related to clustering of
data by various mathematical equations/algorithms. The search
for groupings or patterns in solid-state data is often initiated
whenever the generated data matrix is too large for visual inspection. Evidently, the use of automated HT systems easily performing
thousands of crystallisations requires effective data mining in order to provide a satisfactory overview of all the data in one single
plot. During early screening, it is impossible to predict which specic regions of the diffractogram and/or spectrum (yet undiscovered) the solid forms will affect. Therefore, many of the
unsupervised methods utilise more than just a single peak for
the analysis. HT studies performed using the CrystalMax screen
have demonstrated a simple approach to unsupervised analysis
of Raman data [69,70,88]. Here, spectral similarity was assessed
by calculating the so-called Tanimoto coefcient for each pair-wise
combination of Raman spectra. Tanimoto values close to 0 and 1
indicate high and low similarity, respectively, and the results are
easily visualised using a contour-like plot. In general, the use of
spectral peak positions, for the calculation of some correlation
measure, is a widely accepted approach among researchers involved in HT screening [89]. This approach has also proven useful
for treatment of XRPD data [71]. It is, however, recommended to
show extra care when analysing XRPD data, since changes in crystallinity and preferred orientation can obscure the XRPD pattern
markedly [90]. In these cases, careful ltering of the XRPD data is
imperative to ensure successful data analysis [71].
For treatment of information-rich data, it can be advantageous
to apply multivariate techniques. In multivariate analysis, the measured data are rst organised into a data matrix, having samples/
spectra as rows, and x-variables (wavelength/number, 2h angle,
etc.) as columns. Each combination of sample and x-variable will
have an associated measured value (absorbance, intensity, etc.).
Subsequently, this X-matrix is subjected to various multivariate
techniques, utilising either the whole spectral range (all x-variables) or selected regions. In this respect, cluster analysis [91]
has proven very useful in the search for data clusters in early solid
form screening. This algorithm searches for clusters based on a

29

predened measure of (dis)similarity, typically a distance measure

(e.g., Euclidean, Mahalanobis, city-block) or correlation coefcients. Sample (dis)similarity is hierarchically visualised in a dendrogram. Cluster analysis, with correlation coefcient as
(dis)similarity) measure, has previously been used for analysis of
XRPD [92] and Raman data [87] in HT screening experiments and
for NIR data acquired from a bench-scale polymorph screening of
sulfathiazole [93]. An associated problem with cluster analysis is
to determine where to cut the dendrogram, that is, to estimate
the numbers of actual clusters [59,87]. Therefore, it is highly recommended to supplement the cluster analysis with other data
analysis tools to ensure the correct interpretation of the experimental data. In this respect, Storey et al. supported their cluster
analysis by using multi-dimensional scaling (MDS) plotting [92],
whilst Aaltonen et al. visualised data using the widely popular
multivariate technique, principal component analysis (PCA)
[93,94].
PCA is a (bi)linear modelling approach, where the dimensions of
the X-matrix (corresponding to the number x-variables) are reduced into a fewer set of mutually orthogonal (and thus uncorrelated) variables, termed principal components (PCs), best
describing the systematic variance of the data. Each PC is the product of samples-score vectors and variables-loading vectors. Plotting two score vectors against each other will give the position of
samples (i.e., the rows of the X-matrix) in that respective PC direction, while a plot of loading vectors describes the relationship between the original variables (i.e., columns of the X-matrix) and the
PC direction in question. Combined, the score and loading plots
will provide information on how the samples behave mutually
[95]. The linearity of PCA means that the modelling and subsequent interpretation is relatively straightforward. In the abovementioned study [93], NIR data from sulfathiazole samples were
decomposed into three PCs, explaining 99.4% cumulative variance.
The score plot, showing the position of NIR spectra in these three
dimensions, conrmed the presence of three sulfathiazole polymorphs in the sample set. In comparison, solid form screening of
nitrofurantoin by combined NIRRaman required two PCs (94.2%
explained variance) for total exploration of the spectral information [94] and revealed two anhydrate forms and one hydrate.
Jrgensen et al. monitored the dehydration of erythromycin dihydrate by Raman spectroscopy and PCA [96]. The loading plot provided unique insight into the spectral regions of interest for
dehydration phenomena at different temperatures. Another study
reports the use of PCA for analysing in-situ Raman data from a stable form screening experiment of a disclosed compound [9].
As is the case for all data analysis tools, PCA has its limitations.
As such, PCA is a linear modelling approach, and presence of nonlinearities in the dataset often requires a higher number of PCs for
correct modelling [97]. This was illustrated in a study, where the
goal was to visualise the diversity of database containing 218 solvents [58]. Total exploration of the database in the sense that
variance related to all initially selected variables should be described in the model required ve PCs. Clearly, this is not satisfactory from a presentational viewpoint, only permitting two to three
dimensions. To accommodate this issue, it can be advantageous to
supplement the data analysis with non-linear modelling methods.
Self-organising maps (SOMs) in particular, have proven useful for
visualising multi-dimensional data in one single plot. SOM tting
is a so-called natural learning algorithm, in which the data are projected onto a map of nodes, while still preserving as much of the
topology of the original data as possible. Samples sharing similar
properties will be placed in either the same node or in neighbouring regions of the map. Spraggon et al. tted SOM to image data
obtained from a HT crystallisation screen [98]. The SOM made it
possible to identify groupings of crystals based on their morphology (unsupervised part), and subsequently for assigning new crys-

30

J. Aaltonen et al. / European Journal of Pharmaceutics and Biopharmaceutics 71 (2009) 2337

tals to the SOM (supervised part). When dealing with multivariate

spectral data, it is advisable to reduce the number of x-variables in
the dataset prior to SOM modelling [97] by, for example, PCA or a
genetic algorithm. Ultimately, computational time is decreased,
resulting in an increase of overall modelling efciency. Despite
the apparent advantages of SOM, applications in solid form screening are still rare. The lack of literature on this topic might be due to
the complexity of natural learning algorithms, which often require
several decisions on the settings of algorithmic parameters. For
computation of SOMs, the researcher will have to specify parameters such as node distance measure, map size, map shape, neighbourhood function, learning rate, and number of iterations/
epochs. These can all affect the specic location of samples on
the trained map. Moreover, the random initialisation of nodes
(weight-vectors), and the algorithmic approach taken in natural
learning mean that slightly varying results are sometimes generated between repeated SOM runs on the same dataset. Evaluation
of the quality of the map is therefore crucial. Some previously reported procedures include error estimation either during computation of the SOM [97] or ad-hoc, that is, after computation [58].
2.3.4. Supervised methods
Supervised methods are used for the purpose of predicting a
certain sample property. In short, a supervised method is constructed by correlating one or more dependent x-variables to at
least one independent y-variable using different mathematical approaches. The X-matrix once again comprises measured data, for
example, spectroscopic data, while the y-variable designates the
sample property that is to be predicted in unknown samples.
Supervised methods can roughly be separated into two categories:
(1) quantication and (2) classication. Quantication concerns
the prediction of a quantitative parameter for unknown samples,
usually solid-state purity, while classication aims at assigning unknown samples into user-dened classes. Thus, classication is a
qualitative approach. In the current text, focus will be placed on
the quantication issue, due to the increased activity within this
eld resulting in large volumes of information. The reader is referred to excellent textbook material [91] for more information
on classication as well as studies by Aldridge et al. [99] and
Kogermann et al. [100], in which classication algorithms are applied for API solid form detection. In 2001, Stephenson et al.
[101] provided an extensive review on solid form quantication.
The literature available can be divided into two categories; (1)
in-line quality control during processing and (2) off-line determination of solid-state purity in raw and processed materials or in
the nal product. In general, there is plenty of information available in the literature on the quantication of crystalline and amorphous forms.
Activity and interest have especially increased within the eld
of multivariate calibration applied to spectroscopic data (IR, NIR,
Raman). In this respect, principal component regression (PCR)
and partial least squares (PLS) regression represent useful modelling approaches that utilise either the entire or parts of the spectral
region. PCR is in essence a multiple linear regression (MLR) performed on the PCA score values of the samples, the latter representing the systematic variance of the X-matrix [91]. Hence, the
score values are used for prediction of the desired property (i.e.,
the y-variable) instead of the raw data (intensity values in the case
of spectroscopic data). In contrast, the PLS algorithm originates
from the fact that the score values are not necessarily predictive
for y. Thus, the PLS algorithm nds the direction of maximum variance in X relevant to the target property, Y, thereby creating a new
set of scores predictive for Y [91]. As a result, a PLS model usually
requires a fewer number of components/factors than the corresponding PCR model, and thus may provide a model with better
predictive ability. In practice, however, the difference in perfor-

mance between the two is often small and negligible [91,102].

Indomethacin is a model compound that has undergone much research, concerning the quantication of its solid forms retained in
either physical mixtures or in the nal dosage form. Otsuka and coworkers have demonstrated how MLR and PCR can be used for
multivariate modelling of NIR data of binary mixtures of indomethacin solid forms [103,104]. Same authors used NIR and PCR
to predict the amount of c-indomethacin in tablets containing
two different excipients [105]. The calibrated models utilised
either the whole or parts of the NIR spectral range. However, using
all data points does not necessarily provide optimal models. Instead it is recommended to identify the regions containing the
most information with regard to the property being predicted
(the y-variable). This can greatly improve the accuracy, as well as
robustness, of the calibration models. Other studies demonstrate
the use of PCR on Raman data for quantication of binary mixtures
of polymorphs of carbamazepine [106] and ranitidine HCl [107].
PLS has been widely applied to spectroscopic data for the quantitative determination of API solid forms [102,108111]. Patel and coworkers compared the predictive abilities of MLR and PLS algorithms to a univariate approach [112], the latter utilising inverse
least squares (ILS) regression on a normalised peak. The calibration
was performed on second derivative NIR data of binary mixtures
containing sulfathiazole Forms I and III for the purpose of quantifying the polymorphic ratio between the two. For this particular
dataset, the univariate method provided the lowest predictive error compared to its multivariate counterparts. Same conclusions
were drawn in a comparable study concerning the quantication
of sulfamethoxazole Form I relative to Form II [113]. Evidently,
there are situations in which a simple univariate approach may
work better than multivariate analysis. Other studies demonstrate
the use of univariate analysis for quantication of polymorphic
purity in mixtures by FT-IR [114], and for in-situ monitoring of solvent-mediated phase transformation in a batch crystalliser [115]
and during dissolution testing [79] by Raman spectroscopy. On
an end note, it is relevant to underscore the importance of model
performance testing during calibration. For multivariate models,
this is often achieved using cross validation (internal) and/or by
prediction of an independent test set (external). More information
on this topic can be found in reference [91]. Overall, performance
testing is a crucial part of multivariate model development where
the aim is to obtain reliable predictions of sample properties.
As for unsupervised analysis, non-linear relationship between
sample properties and instrumental response may occur. In these
cases non-linear natural learning algorithms may produce more
accurate models, and thus provide a better alternative to the linear
methods. Articial neural networks (ANNs) for non-linear regression have proven very useful, and may be regarded as the supervised counterpart to the previously discussed self-organising
maps (SOMs). Conceptually, ANNs are based on the biological nervous system. From an algorithmic viewpoint, incoming signals (the
input) are passed to an articial neuron body, also denoted as the
linear learning machine, where they are weighted and summed.
Then they are transformed via a transfer function (or output function) into the outgoing signal (the output). The setting of the
weights during the training phase is essential for the proper function of the network, since they determine the relationship between
input (x-variables, e.g., spectral data) and eventual output (y-variable, target property) [91]. The training phase is therefore a crucial
part of the ANNs development process. ANNs modelling has previously been used for determining solid-state purity of indomethacin
by XRPD [116] and ranitidine hydrochloride by XRPD exclusively
[117] or in conjunction with FTIR [118]. The successful application
on XRPD data is claimed to be due to the non-linear nature of
diffraction intensities as a result of preferred orientation effects.
In spite of this, information on the use of ANNs in solid form

J. Aaltonen et al. / European Journal of Pharmaceutics and Biopharmaceutics 71 (2009) 2337

screening is still sparse. The complexity of the algorithm and the

many user-denable settings are perhaps contributing factors to
this fact. Moreover, in spite of offering good predictive properties,
interpretation of ANNs models is quite difcult [119].
As is evident from the discussion above, the task of analysing
complex solid-state data can be carried out using several unsupervised and supervised techniques. As the techniques are often complementary, it is generally advisable to test more than just a single
method. In this context, it must be emphasised that there is a
wealth of multivariate techniques that can be applied and whose
impact on solid form quantication remains to be elucidated
[91]. Undoubtedly, much more research on this topic is to come.

3. Computational methods for prediction of polymorphs

The accurate prediction of a crystal structure and its potential
polymorphs is a difcult task [120,121]. There are two reasons
for this: rstly, the energies that hold a crystal together are comparatively weak in comparison to intramolecular forces. For example, naphthalene is a solid at room temperature but may be
sublimed when heated; this phase change, the sublimation energy,
costs 72.4 kJ mol1 of energy [122,123]. Naphthalene contains 10
sp2-hybridised carbon atoms bonded to each other, these carboncarbon aromatic bonds have bond energies of the order of
500 kJ mol1 each [124]. The upshot of this very different energy
landscape is that computational modelling can provide an accurate
picture of nuclear coordinates and electronic structure for the molecule naphthalene, but the crystal structure, the intermolecular, is
more challenging. With polymorph prediction this difculty is
compounded because the differing polymorphs one may wish to
discover in silico will have very similar energies. Secondly, the
forces that act on molecules to form crystals, and thus determine
the lattice energy, are diverse and possess subtle directional properties [121,125,126]. The energetics of molecules in crystals are
characterised by a number of terms: (1) Coulombic interactions;
(2) Polarisation; (3) Dispersion; (4) Repulsion [120]. To establish
the lattice energy each of these terms must be evaluated. One
way to model the forces is by considering the atoms of the system
and evaluate the atom-atom interactions using a Buckingham-type
potential [121] such as
1
ERij A expBRij  CR6
ij qi qj Rij

in which the energy between atoms i and j is related to the distance

between the atoms (Rij), the charge on each of the atoms (qi and qj,
respectively) and the parameters A, B and C. The rst two terms relate to the dispersion and repulsion forces and the third term to the
Coulombic interaction. Attempts to improve the modelling of the
lattice energy rely upon getting a quantitatively better picture of
the charge distribution about the molecules and the forces that
act between molecules. For neutral molecules in a crystal, the dominant energy terms in the lattice energy are the dispersion and
repulsion terms which are often 10 times greater than the coulombic term [127].
The prediction of different polymorphs then involves the accurate relative energy determination of the order of 12 kJ mol1 [8].
This may seem like an impossible task and is certainly very difcult, but there are a number of successes in the eld that have
overcome the difculties alluded to above. The prediction of crystal
structures is considered so important that the Cambridge Crystallographic Data Centre (CCDC) has established some blind tests
[128130]. The results of the rst two have been discussed by Price
[121].
Polymorph Predictor (Molecular Simulations Inc.) is a commercial software that allows one to analyse organic compounds, such
as pharmaceuticals, for differing types of polymorphs [131]. This

31

package works by using a Monte Carlo simulated annealing approach, a random search procedure that provides many packing
alternates for the compound of interest. The most likely alternates
(those with lowest energy) may be optimised by lattice energy
minimisation, and these structures ranked in energy terms. The
calculated structures may be validated by prediction of the XRPD
pattern and this may be compared to experimental data [131]. This
method allows each molecule within the unit cell to be adjusted in
the minimisation process. This may be done using force elds that
include energy values associated with bond stretching, bending
and torsional displacement [132]. These force elds are constructed from analysis of structural data for a particular set of compound structures the parameterisation set. This method can be
quite successful if the crystal structure of interest has similar
bonding characteristics to the compound structures from which
the force eld was derived [132]. It is, however, more difcult if
the compounds of interest differ signicantly from the parameterisation set. For example, there are very good force elds for hydrocarbon interactions [132]. These have been used very effectively in
modelling hydrocarbon interactions and structures but their
validity as good force elds in examining aromatic or highly functionalised compounds which are often found in pharmaceuticals,
is less clear [126].
In recent years, two other computational methods have contributed to polymorph prediction and crystal structure modelling of
relevance to pharmaceutical systems. The rst of these is the
DMAREL approach developed by Price and co-workers [121,
133,134]. The strategy adopted [72] is to calculate the structure
of the molecule of interest using an ab initio method, with commercially available software [135]. The charge density of the molecule is then represented by a set of atomic point multipoles, that
is, each nuclei is considered to exert a charge that is not simply
dependent on R1 (as shown in Eq. (1)), but also includes higher
terms, up to R5 [136]. This additional consideration gives a superior Coulombic force eld, and thus, in principle, a more accurate
evaluation of energies. The dispersion and repulsion terms are
provided by parameterised sets [137139]. Having obtained an
accurate model of the charge distribution of the individual molecule, a series of densely packed structures may be obtained, the
densest structures are then optimised with respect to the lattice
energy using the DMAREL algorithm [136]. It is also possible to
predict the phonon modes of crystals and compare these to experimental data, which provides a further test of the validity of the
minimised structures [134,140]. The differing conformations that
a molecule may have in its crystal form over a calculated gas phase
structure have been considered and although not implicitly dealt
with in this method there are search strategies to deal with the
repacking of exible molecules [141].
A number of recent studies highlight the success of the DMAREL
approach. Progesterone is used as an oral contraceptive. It is available in an optically pure form, termed nat-progesterone, which has
two polymorphs [142]. The crystal packing landscape of nat-progesterone in a number of molecular conformations was explored
using the DMAREL strategy [13,143]. As part of this analysis, the
mirror image of nat-progesterone, ent-progesterone, was examined
and it was discovered that a racemic mixture of these could pack
with a lower energy than either enantiomer. This was experimentally veried by crystallisation of the racemate [13,143].
In a detailed polymorph screening study of carbamazepine [72],
it was established that the dimer structural motif that is observed
in the 4 known polymorphs is not the only energetically viable motif. An elongated chain structure was also possible. Further to this,
the chain structure was obtained experimentally by co-crystallising CBZ with its dihydro analogue (DHC) [144]. The crystal
structure of DHC had the requisite chain motif and the co-crystal
contained molecules of both DHC and CBZ with this structure.

32

J. Aaltonen et al. / European Journal of Pharmaceutics and Biopharmaceutics 71 (2009) 2337

The PIXEL method, developed by Gavezzotti, is another method

for crystal structure prediction [120,125127,145148]. In this
method, the molecular charge density of the valence electrons is
calculated by an ab initio method across a three-dimensional volume with discrete points within the volume being electroncharge pixels. The molecule is then represented by nuclear charges
and the electron charge as dened by the pixels within the volume
evaluated. The subsequent lattice energies are then determined
using the nuclei and electron-charge pixels within the volume of
interest. This is a rather appealing strategy because if the charge
densities are correctly calculated then the intermolecular forces
may be determined. Intermolecular Coulomb energies may be calculated from the nucleinuclei, electron pixelpixel and nuclei
electron pixel interactions. The other energies, polarisation and
dispersion and repulsion may also be evaluated, although these require the input of some empirical values. This method is not yet
accessible to researchers [126] but it has been successful in correlating lattice energies to sublimation energies for almost 100 organic crystals [127]. In addition to simple polymorphs, the issues
of differing solvates (or hydrates) and charged species add further
complexity to the prediction of structure. The methods described
above can certainly deal with this additional complexity, in principle, but the problem becomes more difcult because of the addition of more atoms and, more critically, the fact that the solvate
molecules are often smaller than the API and have much shallower
interaction energies. This has recently been discussed by Price
[149].
In summary, the calculation of crystal structures is challenging,
however, progress can be made with careful consideration of the
system of interest and judicious choice of force eld. Furthermore,
a number of new strategies that utilise ab initio calculations, and
thus should be more generally applicable, are in development.
4. Solid form screening an industry perspective
In the last decade, big steps have been taken towards the understanding and control of solid forms of APIs. The widespread interest stems not only from the scientic considerations, but also due
to the recently emerged regulatory and IP aspects in this eld. Both
innovator and generic companies have been trying hard to take
intellectual gains from the discovery of new solid forms. As a result, solid form patent litigations have become a bottleneck for
both sides, which takes many efforts and puts nancial burden
on the companies [3]. This section provides an overview on the
regulatory expectations and IP considerations in solid form screening for both innovator and generic companies.
4.1. Innovator companies
Considering the potential impact of solid forms on product performance, regulatory authorities have put due emphasis on solid
form screening and subsequent monitoring. The regulatory guidances have addressed polymorphism in both new drug applications (NDAs) and abbreviated new drug applications (ANDAs)
[150]. The ICH Q6A covers solid forms arising during the development of new APIs and decision trees have been provided [6]. Part I
of decision tree #4 covers the solid form screening of APIs and their
characterisation. However, the extent to which this screen has to
be carried out is not answered in this guidance and it is up to
the discretion of the pharmaceutical company. Logically, this
decision should be based on the solubility, dose and formulation
characteristics of the API [151]. Other emphasis given in this guidance is that the proper control and monitoring should be placed
whenever multiple solid forms are present, and specication with
respect to polymorphic purity should be incorporated if necessary.
The specication on polymorphic purity is very important if the

API has solubility-limited bioavailability and/or is prone to solidstate transformations during processing and/or storage. Therefore,
once a new API is chosen for potential development, it is imperative to screen for the solid forms it may possess, and to identify
the most suitable one with respect to solubility and stability as
early as possible. However, there is no method that can provide
absolute condence that the ideal solid form has been obtained,
and sometimes the nal form used in the product may indeed arise
at the later stages of development [14,152]. Atorvastatin was initially formulated as an amorphous salt during the development
phase. However, it is reported that during Phase III clinical studies,
the salt crystallised and its properties changed, which compelled
the developer, WarnerLambert, to conduct additional bridging
studies to demonstrate acceptability of the new product relative
to that used in the pivotal registration studies [14]. These kinds
of events are costly, consuming more development time as well
as resources. As such innovator companies are free to choose any
suitable solid form as long as there are no IP issues involved, which
is highly unlikely during early stage as knowledge regarding the
API would normally have remained within the company itself.
The knowledge generated by conducting solid form screening can
provide the innovator company an opportunity to build a patent
portfolio around different forms and therefore a means to enhance
product lifecycle management [153]. A convergence of events negatively affecting pharmaceutical product lifecycle has elevated the
value of solid form IP, thus making comprehensive solid form
screening early in the development very important. Innovator
companies have tried to protect solid forms by patents and have
gained extra years for the product beyond the expiry of the basic
molecule patent. One of the earliest cases of this kind was that of
ranitidine hydrochloride, in which GSK had patent protection for
form II even though the basic molecule patent had expired
[154,155]. Although the generic companies were able to launch
products with form I, the form II patent helped postpone the generic entry. Table 6 shows a few more examples where branded
pharmaceutical products from the innovator companies have patent protection due to solid form(s) beyond the expiry of the basic
molecule patent. These data are taken from the US Food and Drug
Administrations (FDA) Orange Book database and there can be
other solid form patents for these molecules which are not present
in the Orange Book [156]. Innovators are required by FDA regulations to identify patents claiming API, formulation, or specic therapeutic use of the new pharmaceutical in connection with the drug
approval process. The patents covering solid forms are also part of
this requirement. By including patent in the Orange Book, the innovator gets the advantage of 30 months stay if ANDA is led with
paragraph IV certication for any of the listed patents, and thus,
can delay the generic entry. However, the NDA applicant or holder
is required to submit a patent claiming a different polymorph from
that described in the NDA if a product containing the new polymorph will perform the same as the product described in the
NDA with respect to dissolution, solubility, and bioavailability
[157]. This means that for any solid form other than the one used
in the NDA, the innovator has to prove that the new solid form has
the same therapeutic effect by conducting bioequivalence studies.
But even if the innovator does not include patent(s) related to
other solid forms in the Orange Book and does not gain from the
30 months stay, they can still get advantages of patent protection.
In a hypothetical case where the solid form patent is the only
limiting factor for the generic entry, generic rms can launch their
product if they can discover new a solid form which does not have
IP protection and has suitable characteristics for product development. This was the case when Teva found a way around Mercks
patents on its crystalline form of alendronate (the active ingredient
in the blockbuster Fosamax) and was able to launch generic version much earlier [158]. Thus, by patenting a maximum number of

33

J. Aaltonen et al. / European Journal of Pharmaceutics and Biopharmaceutics 71 (2009) 2337

Table 6
List of some molecules having solid form patent(s) listed in the Orange Book [156]
Generic name
Clopidogrel bisulphate

Brand name

Plavix

Basic molecule patent (expiry date)

Solid form patent (expiry date)

Solid form patented

US 4847265 (Nov 17, 2011)

US
US
US
US
US
US
US
US
US
US
US
US
US

Polymorph II

Atorvastatin
Olanzapine

Lipitor
Zyprexa

US 5273995 (Dec 28, 2010)

US 5605897 (Feb 25, 2014)

Fexofenadine

Allegra

US 5578610 (Nov 26, 2013)

Donepezil Hydrochloride

Aricept

US 4895841 (Nov 25, 2010)

Gatioxacin
Ziprasidone hydrochloride
Gabapentin

Zymar
Geodon
Neurontin

US 4980470 (Dec 15, 2009)

US 4831031 (Mar 02, 2012)
US 4024175 (May 17, 1994)

possible solid forms, innovators can block this route, even though
these solid forms would not be used in the pharmaceutical product. The case where an alternative solid form patented by another
company has better characteristics can also have major implications. An example of such is topiramate sodium where J&J ended
up licensing and paying royalties for the trihydrate form developed
and patented by Transform Pharmaceuticals [159]. After the
launch of a dosage form by the innovator, a new solid form can
help development of novel drug delivery methods with different
release proles and routes of administration. This is particularly
important for APIs whose poor solubility is the main hurdle in
the development. The above discussion exemplies the importance
of solid form screening for innovator companies. Extensive solid
form screening can not only provide them with scientic advantages, but also help meet regulatory requirements and maximise returns from drug development by means of IP. It can be inferred
from Table 6 that solid form patents can provide extra protection
anywhere between 1 and 9 years after the basic molecule patent
has expired. In case of IP gain, the timing of ling the form patents
is important which should be after the core new chemical entity
(NCE) patent ling, but before a competitor has the opportunity
to perform solid form screening. The timing of solid form patent ling also depends on the properties of the molecule and other kinds
of patent protection such as the formulation and the method of use
present in the patent portfolio [160]. Hence, for innovator companies it is logical to carry out solid form screening either before clinical trials or early in clinical trials to maximise the benet to drug
development and reduce IP risk.
4.2. Generic companies
W.C. McCrones statement regarding polymorphism, the number of forms known for a given compound is proportional to the time
and energy spent in research on that compound, becomes quite evident when looking at the contribution of generic pharmaceutical
companies in the discovery of new solid forms of APIs. Once the
branded pharmaceutical product reaches the market and its API
has shown business potential, generic companies start putting effort into solid form screening. Compared to the innovator companies, the generic companies are able to invest more in chemical
development and process chemistry simply because they can rely
on much of the other work already done by the innovator. As innovators are protecting more and more solid forms by patents, solid
form screening by generic companies becomes imperative for an
early as possible launch of a generic product. For a single branded
pharmaceutical product there are many generic players working
around. Hence, time and energy spent in solid form screening becomes manifold, fullling the stated requirement for the discovery
of new forms. The aspect of polymorphism in generic product
development is well covered in recent publications and regulatory

6429210 (Jun 10, 2019)

6504030 (Jun 10, 2019)
5969156 (Jul 08, 2016)
5736541 (Mar 24, 2015)
6251895 (Sep 23, 2017)
7135571 (May 18, 2014)
7138524 (May 18, 2014)
5985864 (Dec 30, 2016)
6140321 (Dec 30, 2016)
6245911 (Dec 01, 2018)
5880283 (Dec 05, 2015)
5312925 (Sep 01, 2012)
4894476 (May 02, 2008)

Polymorph I, II, IV and their hydrates

Polymorph II,
Dihydrate D
Anhydrate Form I,
Hydrate Form II
Polymorph II, III,
IV, V and A, B, C
Sesquihydrate
Monohydrate
Monohydrate

guidances [150,161,162]. FDA issued its latest guidance on polymorphism-related issues to be considered for ANDA submission
in 2007 [150]. This guidance is more elaborate compared to ICH
Q6A and provides decision trees for the development of a generic
product. Decision tree #1 provides recommendations on when
specications for polymorphic form(s) for the API and/or the pharmaceutical product may be appropriate. Polymorphs are unlikely
to have a signicant effect on bioavailability when all the forms
have the same apparent solubilities or all the forms are highly soluble. The guidance puts emphasis on solid form screening to get
knowledge about all the solid forms that the API may have along
with the use of published literature and patents. Decision tree #2
gives an approach for setting specications for polymorphs when
at least one form is known to have low solubility based on the
BCS. Decision tree #3 provides an approach when considering
whether to set specications for polymorphs in the pharmaceutical
product. Generally, specications for polymorphs in pharmaceutical products are not necessary if the most thermodynamically stable polymorphic form is used or if the same form is used in an
approved product of the same dosage form. The recommendations
regarding which solid forms to be considered for monitoring and
control are given based on the physicochemical properties of the
API. Moreover, it has claried the issue of sameness in ANDAs,
in which the guidance mentions that the different polymorphic
forms do not render the substances different APIs for the purpose
of ANDA approvals. Here, the term polymorphic forms includes
polymorphs, amorphous form, solvates and hydrates. Over the
years, FDA has approved many ANDAs having the API in a different
solid form from the one in the respective Reference Listed Drug
(RLD). Therefore, if a solid form present in a RLD is still under patent protection and the basic molecule patent has expired, generic
companies can develop their products with another solid form,
provided it meets other requirements of FDA for ANDA approval.
Generic companies are working hard to tap such opportunities,
and competition has forced them to carry out solid form screening
as early as possible. In the US, the rst ANDA approved with paragraph IV certication (i.e., particular Orange Book patent(s) is invalid, unenforceable, or will not be infringed) is entitled to 180 days
marketing exclusivity [163]. In many cases where the solid form
patent is listed in the Orange Book, generic companies have opted
to le ANDA with a new solid form developed by their own solid
form screening which provides an opportunity for paragraph IV
certication. So by carrying out thorough screening, generic companies can not only develop their own generic products with
new solid forms, but they can also block other generic launches
with the help of marketing exclusivity and patent protection.
One notable example here is sertraline hydrochloride, which is
the active ingredient in the blockbuster antidepressant Zoloft.
Teva led ANDA with paragraph IV certication on Orange Book
listed patent US 5248699 which claims sertraline polymorph, and

34

J. Aaltonen et al. / European Journal of Pharmaceutics and Biopharmaceutics 71 (2009) 2337

eventually gained 180 days marketing exclusivity. Teva secured

patent protection on its novel solid forms by several patents and
later went on to sue many other generic companies who had led
ANDAs for sertraline hydrochloride. In fact, this race of being the
rst generic has created a situation where there are more solid
forms patented by generic companies compared to innovators.
Overall, it is prudent for the generic companies to carry out extensive solid form screening to get the best possible advantages from
new solid forms.
5. Conclusions
The development of a new pharmaceutical requires a deep
understanding of solid-state phenomena. In order for a pharmaceutical product to succeed, the possible existence and performance of various solid forms need to be thoroughly investigated
and the IP strategies carefully considered before commercial
launch. Therefore, solid form screening has become an essential
part of pharmaceutical development and product lifecycle
management.
The number of new pharmaceuticals is decreasing, and the molecules that reach the later phases of drug development are becoming ever more structurally complex, which creates further
challenges for solid form screening. New biomacromolecular compounds are often administered as solutions. However, the solid
state is, and will remain, the most stable state of matter and for
this reason, a very attractive option for formulation. In the future,
research should focus on not only optimisation of the crystalline
material, but also exploring the means for stabilisation of amorphous formulations for increased solubility, and thus
bioavailability.
The tools available for solid form screening have evolved radically during the past decade we can now explore the solid forms
computationally, perform thousands of experimental crystallisations with miniaturised high-throughput screening technologies,
and identify new solid phases fast. We should, however, pay special attention on the evaluation part of the screening. Understanding of basic thermodynamics together with robust design of
experiments and powerful data analysis are the keys to successful
solid form screening.
Acknowledgement
The Academy of Finland, Finnish Cultural Foundation and Finnish Pharmaceutical Society are acknowledged for funding (J.A.).
Professor Thomas Rades is thanked for fruitful discussions.
References
[1] J. Haleblian, W. McCrone, Pharmaceutical applications of polymorphism, J.
Pharm. Sci. 58 (1969) 911929.
[2] J. Thun, L. Seyfarth, J. Senker, R.E. Dinnebier, J. Breu, Polymorphism in
Benzamide: solving a 175-year-old riddle, Angew. Chem., Int. Ed. 46 (2007)
67296731.
[3] J. Bernstein, Polymorphism and patents, Polymorphism in Molecular Crystals,
Clarendon Press, Oxford, 2002. pp. 297307.
[4] J. Bulow, The gaming of pharmaceutical patents, in: A.B. Jaffe, J. Lerner, S.
Stern (Eds.), Innovation Policy and the Economy 4, The MIT Press, Cambridge,
MA, 2004, pp. 145185.
[5] S.R. Chemburkar, J. Bauer, K. Deming, H. Spiwek, K. Patel, J. Morris, R. Henry, S.
Spanton, W. Dziki, W. Porter, J. Quick, P. Bauer, J. Donaubauer, B.A. Narayanan,
M. Soldani, D. Riley, K. McFarland, Dealing with the impact of Ritonavir
polymorphs on the late stages of bulk drug process development, Org. Process
Res. Dev. 4 (2000) 413417.
[6] ICH Harmonised Tripartite Guideline Q6A, Test procedures and acceptance
criteria for new drug substances and new drug products: chemical
substances, International Conference on Harmonisation, 1999.
[7] R.J. Davey, Crystallization: how come you look so good?, Nature 428 (2004)
374375
[8] J.D. Dunitz, Are crystal structures predictable?, Chem Commun. (2003) 545
548.

[9] J.M. Miller, B.M. Collman, L.R. Greene, D.J.W. Grant, A.C. Blackburn, Identifying
the stable polymorph early in the drug discovery-development process,
Pharm. Dev. Tech. 10 (2005) 291297.
[10] C.-H. Gu, V. Young, D.W.J. Grant, Polymorph screening: inuence of solvents
on the rate of solvent-mediated polymorphic transformation, J. Pharm. Sci. 90
(2001) 18781890.
[11] D. Singhal, W. Curatolo, Drug polymorphism and dosage form design: a
practical perspective, Adv. Drug Del. Rev. 56 (2004) 335347.
[12] W.I. Cross, N. Blagden, R.J. Davey, R.G. Pritchard, M.A. Neumann, R.J. Roberts,
R.C. Rowe, A whole output strategy for polymorph screening: combining
crystal structure prediction, graph set analysis, and targeted crystallization
experiments in the case of diunisal, Cryst. Growth Des. 3 (2003) 151158.
[13] R.W. Lancaster, P.G. Karamertzanis, A.T. Hulme, D.A. Tocher, D.F. Covey, S.L.
Price, Racemic progesterone: predicted in silico and produced in the solid
state, Chem. Commun. (2006) 49214923.
[14] C.R. Gardner, C.T. Walsh, . Almarsson, Drugs as materials: valuing physical
form in drug discovery, Nat. Rev. Drug Discov. 3 (2004) 926934.
[15] S.R. Byrn, R.R. Pfeiffer, J.G. Stowell, Polymorphs, Solid-State Chemistry of
Drugs, SSCI Inc., West Lafayette, 1999. pp. 143231.
[16] D.W.J. Grant, Theory and origin of polymorphism, in: H.G. Brittain (Ed.),
Polymorphism in Pharmaceutical Solids, Marcel Dekker Inc., New York, 1999,
pp. 133.
[17] E.V. Boldyreva, T.P. Shakhtshneider, H. Ahsbahs, H. Sowa, H. Uchtmann, Effect
of high pressure on the polymorphs of paracetamol, J. Therm. Anal. Calorim.
68 (2002) 437452.
[18] R.J. Davey, N. Blagden, G.D. Potts, R. Docherty, Polymorphism in molecular
crystals: stabilization of a metastable form by conformational mimicry, J. Am.
Chem. Soc. 119 (1997) 17671772.
[19] P. Vishweshwar, J.A. McMahon, J.A. Bis, M.J. Zaworotko, Pharmaceutical cocrystals, J. Pharm. Sci. 95 (2006) 499516.
[20] S.R. Byrn, R.R. Pfeiffer, J.G. Stowell, Hydrates and solvates, Solid-State
Chemistry of Drugs, SSCI Inc., West Lafayette, 1999.
[21] A.L. Gillon, N. Feeder, R.J. Davey, R. Storey, Hydration in molecular crystals a
Cambridge Structural Database analysis, Cryst. Growth Des. 3 (2003) 663
673.
[22] K.R. Morris, Structural aspects of hydrates and solvates, in: H.G. Brittain (Ed.),
Polymorphism in Pharmaceutical Solids, Marcel Dekker Inc., New York, 1999,
pp. 125181.
[23] B.C. Hancock, G. Zogra, Characteristics and signicance of the amorphous
state in pharmaceutical systems, J. Pharm. Sci. 86 (1997) 112.
[24] N. Blagden, R.J. Davey, R. Rowe, R. Roberts, Disappearing polymorphs and the
role of reaction by-products: the case of sulphathiazole, Int. J. Pharm. 172
(1998) 169177.
[25] L. Yu, Amorphous pharmaceutical solids: preparation, characterization and
stabilization, Adv. Drug Del. Rev. 48 (2001) 2742.
[26] A.M. Kaushal, P. Gupta, A.K. Bansal, Amorphous drug delivery systems:
molecular aspects, design, and performance, Crit. Rev. Ther. Drug Carrier Syst.
21 (2004) 133193.
[27] B.C. Hancock, M. Parks, What is the true solubility advantage for amorphous
pharmaceuticals?, Pharm Res. 17 (2000) 397404.
[28] T.L. Threlfall, Analysis of organic polymorphs. A review, Analyst 120 (1995)
24352460.
[29] G.L. Amidon, H. Lennerns, V.P. Shah, J.R. Crison, A theoretical basis for a
biopharmaceutic drug classication: the correlation of in vitro drug product
dissolution and in vivo bioavailability, Pharm. Res. 12 (1995) 413420.
[30] P. Kahela, R. Aaltonen, E. Lewing, M. Anttila, E. Kristoffersson,
Pharmacokinetics and dissolution of two crystalline forms of
carbamazepine, Int. J. Pharm. 14 (1983) 103112.
[31] M.C. Meyer, A.B. Straughn, E.J. Jarvi, G.C. Wood, F.R. Pelsor, V.P. Shah, The
bioinequivalence of carbamazepine tablets with a history of clinical failures,
Pharm. Res. 9 (1992) 16121616.
[32] P. Di Martino, A.M. Guyot-Hermann, P. Conant, M. Drache, J.C. Guyot, A new
pure paracetamol for direct compression: the orthorhombic form, Int. J.
Pharm. 128 (1996) 18.
[33] G. Nichols, C.S. Frampton, Physicochemical characterization of the
orthorhombic polymorph of paracetamol crystallized from solution, J.
Pharm. Sci. 87 (1998) 684693.
[34] E. Suihko, V.-P. Lehto, J. Ketolainen, E. Laine, P. Paronen, Dynamic solid-state
and tabletting properties of four theophylline forms, Int. J. Pharm. 217 (2001)
225236.
[35] C. Sun, D.J.W. Grant, Inuence of crystal structure on the tableting properties
of sulfamerazine polymorphs, Pharm. Res. 18 (2001) 274280.
[36] S.R. Byrn, W. Xu, A.W. Newman, Chemical reactivity in solid-state
pharmaceuticals: formulation implications, Adv. Drug Del. Rev. 48 (2001)
115136.
[37] S.R. Byrn, P.A. Sutton, B. Tobias, J. Frye, P. Main, Crystal structure, solid-state
NMR spectra, and oxygen reactivity of ve crystal forms of prednisolone tertbutylacetate, J. Am. Chem. Soc. 110 (1988) 16091614.
[38] Y. Guo, S.R. Byrn, G. Zogra, Physical characteristics and chemical degradation
of amorphous quinapril hydrochloride, J. Pharm.Sci. 89 (2000) 128143.
[39] R.J. Davey, J. Garside, From Molecules to Crystallizers. An Introduction to
Crystallization, Oxford Science Publications, Oxford, 1999.
[40] J.W. Mullin, Industrial techniques and equipment, in: Crystallization,
Butterworth Heinemann, Oxford, 2001, pp. 315402.
[41] M. Volmer, Die chemische reaktion. Bd. IV. Kinetik der phasenbildung,
Steinkopf, Leipzig, 1939.

J. Aaltonen et al. / European Journal of Pharmaceutics and Biopharmaceutics 71 (2009) 2337

[42] S. Khoskhoo, J. Anwar, Crystallization of polymorphs: the effect of solvent, J.
Phys. D: Appl. Phys. 26 (1993) 9093.
[43] J. Bernstein, R.J. Davey, J.-O. Henck, Concominant polymorphs, Angew. Chem.,
Int. Ed. 38 (1999) 34403461.
[44] R. Hilker, S.M. De Paul, M. Szelagiewcz, Approaches to polymorphism
screening, in: R. Hilker (Ed.), Polymorphism in the Pharmaceutical Industry,
Wiley-VCH Verlag GmbH, Weinheim, 2006, pp. 287308.
[45] J.K. Guillory, Generation of polymorphs, hydrates, solvates and amorphous
solids, in: H.G. Brittain (Ed.), Polymorphism in Pharmaceutical Solids, Marcel
Dekker Inc., New York, 1999, pp. 183226.
[46] S.L. Morissette, . Almarsson, M.L. Peterson, J.F. Remenar, M.J. Read, A.V.
Lemmo, S. Ellis, M.J. Cima, C.R. Gardner, High-throughput crystallization:
polymorphs, salts, co-crystals and solvates of pharmaceutical solids, Adv.
Drug Del. Rev. 56 (2004) 275300.
[47] S. Mirza, I. Miroshnyk, J. Heinmki, L. Christiansen, M. Karjalainen, J. Yliruusi,
Inuence of solvents on the variety of crystalline forms of erythromycin,
AAPS PharmSci 5 (2003) (Article 12).
[48] I. Miroshnyk, L. Khriachtchev, S. Mirza, J. Rantanen, J. Heinmki, J. Yliruusi,
Insight into thermally induced phase transformations of erythromycin A
dihydrate, Cryst. Growth Des. 6 (2006) 369374.
[49] S.L. Childs, L.J. Chyall, J.T. Dunlap, D.A. Coates, B.C. Stahly, G.P. Stahly, A
metastable polymorph of metformin hydrochloride: isolation and
characterization using capillary crystallization and thermal microscopy
techniques, Cryst. Growth Des. (2004) 441449.
[50] A.S. Myerson, B.A. Garetz, Method for using laser light to control crystal form,
U.S. Patent 6426406, 2002.
[51] G. Ruecroft, D. Hipkiss, T. Ly, N. Maxted, P.W. Cains, Sonocrystallization: the
use of ultrasound for improved industrial crystallization, Org. Process Res.
Dev. 9 (2005) 923932.
[52] N. Rodrguez-Hornedo, S.J. Nehm, A. Jayasankar, Cocrystals: design properties
and formation mechanisms, in: J. Swarbrick (Ed.), Encyclopedia of
Pharmaceutical Technology, third ed., Informa Healthcare USA, Inc., New
York, 2007, pp. 615635.
[53] G.P. Stahly, Diversity in single- and multiple-component crystals. The search
for and prevalence of polymorphs and cocrystals, Cryst. Growth Des. 7 (2007)
10071026.
[54] B.Y. Shekunov, P. York, Crystallization processes in pharmaceutical
technology and drug delivery design, J. Cryst. Growth 211 (2000) 122136.
[55] M. Kitamura, K. Nakamura, Effects of solvent composition and temperature
on polymorphism and crystallization behavior of thiazole-derivative, J. Cryst.
Growth 236 (2002) 676686.
[56] M. Kitamura, Controlling factor of polymorphism in crystallization process, J.
Cryst. Growth 237239 (2002) 22052214.
[57] C.-H. Gu, H. Li, R.B. Gandhi, K. Raghavan, Grouping solvents by statistical
analysis of solvent property parameters: implication to polymorph screening,
Int. J. Pharm. 283 (2004) 117125.
[58] M. Alles, F. van den Berg, C. Cornett, F.S. Jrgensen, B. Halling-Srensen, H.
Lopez de Diego, L. Hovgaard, J. Aaltonen, J. Rantanen, Solvent diversity in
polymorph screening, J. Pharm. Sci. 97 (2008) 21452159.
[59] M. Alles, J. Rantanen, J. Aaltonen, C. Cornett, F. van den Berg, Solvent subset
selection for polymorph screening, J. Chemom., in press, doi:10.1002/
cem.1107.
[60] D. Xu, N. Redman-Furey, Statistical cluster analysis of pharmaceutical
solvents, Int. J. Pharm. 339 (2007) 175188.
[61] L.-F. Huang, W.-Q. Tong, Impact of solid state properties on
developability assessment of drug candidates, Adv. Drug Del. Rev. 56
(2004) 321334.
[62] W. Beckmann, Seeding the desired polymorph: background, possibilities,
limitations, and case studies, Org. Process Res. Dev. 4 (2000) 372383.
[63] C.-H. Gu, K. Chatterjee, V. Young, D.J.W. Grant, Stabilization of a metastable
polymorph of sulfamerazine by structurally related additives, J. Cryst. Growth
235 (2002) 471481.
[64] X. He, J.G. Stowell, K.R. Morris, R.R. Pfeiffer, H. Li, G.P. Stahly, S.R. Byrn,
Stabilization of a metastable polymorph of 4-methyl-2-nitroacetanilide by
isomorphic additives, Cryst. Growth Des. 1 (2001) 305312.
[65] A.L. Grzesiak, M. Lang, K. Kim, A.J. Matzger, Comparison of the four anhydrous
polymorphs of carbamazepine and the crystal structure of form I, J. Pharm.
Sci. 92 (2003) 22602271.
[66] M. Lang, J.W. Kampf, A.J. Matzger, Form IV of carbamazepine, J. Pharm. Sci. 91
(2002) 11861190.
[67] A. Park, L.J. Chyall, J. Dunlap, C. Schertz, D. Jonaitis, B.C. Stahly, S. Bates, R.
Shipplett, S. Childs, New solid-state chemistry technologies to bring better
drugs to market: knowledge-based decision making, Expert Opin. Drug
Discov. 2 (2007) 145154.
[68] P.J. Desrosiers, The potential of preform, Mod. Drug Discov. (2004) 4043.
[69] S.L. Morissette, S. Soukasene, D. Levinson, M.J. Cima, . Almarsson,
Elucidation of crystal form diversity of the HIV protease inhibitor Ritonavir
by high-throughput crystallization, Proc. Natl. Acad. Sci. USA 100 (2003)
21802184.
[70] M.L. Peterson, S.L. Morissette, C. McNulty, A. Goldsweig, P. Shaw, M.
LeQuesne, J. Monagle, N. Encina, J. Marchionna, A. Johnson, J. GonzalezZugasti, A.V. Lemmo, S.J. Ellis, M.J. Cima, . Almarsson, Iterative highthroughput polymorphism studies on acetaminophen and an experimentally
derived structure for form III, J. Am. Chem. Soc. 124 (2002) 1095810959.
[71] . Almarsson, M.B. Hickey, M.L. Peterson, S.L. Morissette, S. Soukasene, C.
McNulty, M. Tawa, J.M. MacPhee, J.F. Remenar, High-throughput surveys of

[72]

[73]
[74]
[75]
[76]
[77]

[78]

[79]

[80]

[81]
[82]

[83]

[84]

[85]

[86]

[87]

[88]

[89]

[90]
[91]

[92]

[93]

[94]

[95]
[96]

[97]
[98]
[99]

[100]

35

crystal form diversity of highly polymorphic pharmaceutical compounds,

Cryst. Growth Des. 3 (2003) 927933.
A.J. Florence, A. Johnston, S.L. Price, H. Nowell, A.R. Kennedy, N. Shankland, An
automated parallel crystallization search for predicted crystal structures and
packing motifs of carbamazepine, J. Pharm. Sci. 95 (2006) 19181930.
H.G. Brittain, Physical Characterization of Pharmaceutical Solids, Marcel
Dekker Inc., New York, 1995.
R. Hilker, Polymorphism: In the Pharmaceutical Industry, Wiley-VCH Verlag
GmbH KGaA, Weinheim, 2006.
H.G. Brittain, Polymorphism in Pharmaceutical Solids, Marcel Dekker Inc.,
New York, 1999.
D. Giron, M. Mutz, S. Garnier, Solid-state of pharmaceutical compounds, J.
Thermal Anal. Calorim. 77 (2004) 709747.
R.A. Lane, G. Buckton, The novel combination of dynamic vapour sorption
gravimetric analysis and near infra-red spectroscopy as a hyphenated
technique, Int. J. Pharm. 207 (2000) 4956.
A.D. Gift, L.S. Taylor, Hyphenation of Raman spectroscopy with gravimetric
analysis to interrogate watersolid interactions in pharmaceutical systems, J.
Pharm. Biomed. Anal. 43 (2007) 1423.
J. Aaltonen, P. Heinnen, L. Peltonen, H. Kortejrvi, V.P. Tanninen, L.
Christiansen, J. Hirvonen, J. Yliruusi, J. Rantanen, In situ measurement of
solvent-mediated phase transformations during dissolution testing, J. Pharm.
Sci. 95 (2006) 27302737.
A.M. Dyas, U.U. Shah, Dissolution and dissolution testing, in: J. Swarbrick
(Ed.), Encyclopedia of Pharmaceutical Technology, third ed., vol. 2, Informa
Healthcare USA, Inc., New York, 2007.
D.C. Montgomery, Design and Analysis of Experiments, Wiley, New York,
2001.
L.X. Yu, R.A. Lionberger, A.S. Raw, R. DCosta, Applications of process
analytical technology to crystallization process, Adv. Drug Del. Rev. 56
(2004) 349369.
N. Al-Zoubi, K. Kachrimanis, S. Malamataris, Effects of harvesting and cooling
on crystallization and transformation of orthorhombic paracetamol in
ethanolic solution, Eur. J. Pharm. Sci. 17 (2002) 1321.
M. Sheikhzadeh, S. Murad, S. Rohani, Response surface analysis of solutionmediated polymorphic transformation of buspirone hydrochloride, J. Pharm.
Biomed. Anal. 45 (2007) 227236.
M. Fujiwara, Z.K. Nagy, J.W. Chew, R.D. Braatz, First-principles and direct
design approaches for the control of pharmaceutical crystallization, J. Process
Contr. 15 (2005) 493504.
R. Bro, M. Jakobsen, Exploring complex interactions in designed data using
GEMANOVA. Color changes in fresh beef during storage, J. Chemom. 16
(2002) 294304.
S. Lowry, D. Dalrymple, A. Song, V. Rosso, C. Pommier, J. Venit, Integrating a
Raman microscope into the workow of a high-throughput crystallization
laboratory, J. Ass. Lab. Automat. 11 (2006) 7584.
J.F. Remenar, J.M. MacPhee, B.K. Larson, V.A. Tyagi, J.H. Ho, D.A. McIlroy, M.B.
Hickey, P.B. Shaw, O. Almarsson, Salt selection and simultaneous
polymorphism assessment via high-throughput crystallization: the case of
sertraline, Org. Process Res. Dev. 7 (2003) 990996.
R. Hilker, J. Berghausen, F. Blatter, A. Burkhard, S. De Paul, B. Freiermuth, A.
Geoffroy, U. Hofmeier, C. Marcolli, B. Siebenhaar, M. Szelagiewicz, A. Vit, M.
von Raumer, Polymorphism integrated approach from high-throughput
screening to crystallization optimization, J. Thermal Anal. Calorim. 73 (2003)
429440.
S. Datta, D.W.J. Grant, Crystal structures of drugs: advances in determination,
prediction and engineering, Nat. Rev. Drug Discov. 3 (2004) 4257.
B.G.M. Vandeginste, D.L. Massart, L.M.C. Buydens, S. De Jong, P.J. Lewi, J.
Smeyers-Verbeke, Handbook of Chemometrics and Qualimetrics Part B,
Elsevier, Amsterdam, 1998.
R. Storey, R. Docherty, P. Higginson, C. Dallman, C. Gilmore, G. Barr, W. Dong,
Automation of solid form screening procedures in the pharmaceutical
industry how to avoid the bottlenecks, Crystallogr. Rev. 10 (2004) 4556.
J. Aaltonen, J. Rantanen, S. Siiria, M. Karjalainen, A. Jorgensen, N. Laitinen, M.
Savolainen, P. Seitavuopio, M. Louhi-Kultanen, J. Yliruusi, Polymorph
screening using near-infrared spectroscopy, Anal. Chem. 75 (2003) 5267
5273.
J. Aaltonen, C.J. Strachan, K. Pollanen, J. Yliruusi, J. Rantanen, Hyphenated
spectroscopy as a polymorph screening tool, J. Pharm. Biomed. Anal. 44
(2007) 477483.
S. Wold, K. Esbensen, P. Geladi, Principal component analysis, Chemom. Intell.
Lab. Syst. 2 (1987) 3752.
A.C. Jrgensen, I. Miroshnyk, M. Karjalainen, K. Jouppila, S. Siiria, O.
Antikainen, J. Rantanen, Multivariate data analysis as a fast tool in
evaluation of solid state phenomena, J. Pharm. Sci. 95 (2006) 906916.
B.K. Lavine, C.E. Davidson, D.J. Westover, Spectral pattern recognition using
self-organizing maps, J. Chem. Inf. Comp. Sci. 44 (2004) 10561064.
G. Spraggon, S.A. Lesley, A. Kreusch, J.P. Priestle, Computational analysis of
crystallization trials, Acta Crystallogr., Sect. D 58 (2002) 19151923.
P.K. Aldridge, C.L. Evans, H.W.I. Ward, S.T. Colgan, N. Boyer, P.J. Gemperline,
Near-IR detection of polymorphism and process-related substances, Anal.
Chem. 68 (1996) 9971002.
K. Kogermann, J. Aaltonen, C.J. Strachan, K. Pollnen, P. Veski, J. Heinmki, J.
Yliruusi, J. Rantanen, Qualitative in situ analysis of multiple solid-state forms
using spectroscopy and partial least squares discriminant modeling, J. Pharm.
Sci. 96 (2007) 18021820.

36

J. Aaltonen et al. / European Journal of Pharmaceutics and Biopharmaceutics 71 (2009) 2337

[101] G.A. Stephenson, R.A. Forbes, S.M. Reutzel-Edens, Characterization of the

solid state: quantitative issues, Adv. Drug Del. Rev. 48 (2001) 6790.
[102] H.Y. Aboul-Enein, A.A. Bunaciu, S. Fleschin, Analysis of mebendazole
polymorphs by Fourier transform IR spectrometry using chemometric
methods, Biopolymers 67 (2002) 5660.
[103] M. Otsuka, F. Kato, Y. Matsuda, Comparative evaluation of the degree of
indomethacin crystallinity by chemoinfometrical Fourier-transformed nearinfrared spectroscopy and conventional powder X-ray diffractometry, AAPS
PharmSci 2 (2000) (Article 9).
[104] M. Otsuka, F. Kato, Y. Matsuda, Determination of indomethacin polymorphic
contents by chemometric near-infrared spectroscopy and conventional
powder X-ray diffractometry, Analyst 126 (2001) 15781582.
[105] M. Otsuka, F. Kato, Y. Matsuda, Y. Ozaki, Comparative determination of
polymorphs of indomethacin in powders and tablets by chemometrical nearinfrared spectroscopy and X-ray powder diffractometry, AAPS PharmSciTech
4 (2003) (Article 19).
[106] C.J. Strachan, D. Pratiwi, K.C. Gordon, T. Rades, Quantitative analysis
of polymorphic mixtures of carbamazepine by Raman spectroscopy
and principal components analysis, J. Raman Spectrosc. 35 (2004)
347352.
[107] D. Pratiwi, J.P. Fawcett, K.C. Gordon, T. Rades, Quantitative analysis of
polymorphic mixtures of ranitidine hydrochloride by Raman spectroscopy
and principal components analysis, Eur. J. Pharm. Biopharm. 54 (2002) 337
341.
[108] J.W. Braga, R.J. Poppi, Figures of merit for the determination of the
polymorphic purity of carbamazepine by infrared spectroscopy and
multivariate calibration, J. Pharm. Sci. 93 (2004) 21242134.
[109] K.R. Lee, G. Zuber, L. Katrincic, Chemometrics approach to the determination
of polymorphism of a drug compound by infrared spectroscopy, Drug Dev.
Ind. Pharm. 26 (2000) 135147.
[110] K. Pllnen, A. Hkkinen, M. Huhtanen, S.P. Reinikainen, M. Karjalainen, J.
Rantanen, M. Louhi-Kultanen, L. Nystrm, DRIFT-IR for quantitative
characterization of polymorphic composition of sulfathiazole, Anal. Chim.
Acta 544 (2005) 108117.
[111] A. Heinz, M. Savolainen, T. Rades, C. Strachan, Quantifying ternary mixtures
of different solid-state forms of indomethacin by Raman and near-infrared
spectroscopy, Eur. J. Pharm. Sci. 32 (2007) 182192.
[112] A.D. Patel, P.E. Luner, M.S. Kemper, Low-level determination of polymorph
composition in physical mixtures by near-infrared reectance spectroscopy,
J. Pharm. Sci. 90 (2001) 360370.
[113] A.D. Patel, P.E. Luner, M.S. Kemper, Quantitative analysis of polymorphs in
binary and multi-component powder mixtures by near-infrared reectance
spectroscopy, Int. J. Pharm. 206 (2000) 6374.
[114] R. Helmy, G.X. Zhou, Y.W. Chen, L. Crocker, T. Wang, R.M. Wenslow Jr., A.
Vailaya, Characterization and quantitation of aprepitant drug substance
polymorphs by attenuated total reectance fourier transform infrared
spectroscopy, Anal. Chem. 75 (2003) 605611.
[115] J. Schll, D. Bonalumi, L. Vicum, M. Mazzotti, M. Muller, In situ monitoring
and modeling of the solvent-mediated polymorphic transformation of Lglutamic acid, Cryst. Growth Des. 6 (2006) 881891.
[116] T. Okumura, M. Nakazono, M. Otsuka, K. Takayama, An accurate quantitative
analysis of polymorphs based on articial neural networks, Colloids Surf. B 49
(2006) 153157.
[117] S. Agatonovic-Kustrin, V. Wu, T. Rades, D. Saville, I.G. Tucker, Powder
diffractometric assay of two polymorphic forms of ranitidine hydrochloride,
Int. J. Pharm. 184 (1999) 107114.
[118] S. Agatonovic-Kustrin, T. Rades, V. Wu, D. Saville, I.G. Tucker, Determination
of polymorphic forms of ranitidine-HCl by DRIFTS and XRPD, J. Pharm.
Biomed. Anal. 25 (2001) 741750.
[119] D.L. Massart, B.G.M. Vandeginste, L.M.C. Buydens, S. De Jong, P.J. Lewi, J.
Smeyers-Verbeke, Handbook of Chemometrics and Qualimetrics Part A,
Elsevier, Amsterdam, 1998.
[120] A. Gavezzotti, Ten years of experience in polymorph prediction: what next?,
CrystEngComm 4 (2002) 343347
[121] S.L. Price, The computational prediction of pharmaceutical crystal structures
and polymorphism, Adv. Drug Del. Rev. 56 (2004) 301319.
[122] D. Williams, Coulombic interactions in crystalline hydrocarbons, Acta
Crstallogr. Sect. A 30 (1974) 7177.
[123] R.S. Bradley, T.G. Cleasby, The vapour pressure and lattice energy of some
aromatic ring compounds, J. Chem. Soc. (1953) 16901692.
[124] P.W. Atkins, Physical Chemistry, Oxford University Press, Oxford, 1992.
[125] A. Gavezzotti, A solid-state chemists view of the crystal polymorphism of
organic compounds, J. Pharm. Sci. 96 (2007) 22322241.
[126] J.D. Dunitz, A. Gavezzotti, Molecular recognition in organic crystals: directed
intermolecular bonds or nonlocalized bonding?, Angew Chem., Int. Ed. 44
(2005) 17661787.
[127] A. Gavezzotti, Quantitative ranking of crystal packing modes by systematic
calculations on potential energies and vibrational amplitudes of molecular
dimers, J. Chem. Theory Comput. 1 (2005) 834840.
[128] W.D.S. Motherwell, H.L. Ammon, J.D. Dunitz, A. Dzyabchenko, P. Erk, A.
Gavezzotti, D.W.M. Hofmann, F.J.J. Leusen, J.P.M. Lommerse, W.T.M. Mooij,
S.L. Price, H. Scheraga, B. Schweizer, M.U. Schmidt, B.P. van Eijck, P. Verwer,
D.E. Williams, Crystal structure prediction of small organic molecules: a
second blind test, Acta Crystallogr. Sect. B 58 (2002) 647661.
[129] J.P.M. Lommerse, W.D.S. Motherwell, H.L. Ammon, J.D. Dunitz, A. Gavezzotti,
D.W.M. Hofmann, F.J.J. Leusen, W.T.M. Mooij, S.L. Price, B. Schweizer, M.U.

[130]

[131]
[132]
[133]

[134]

[135]

[136]

[137]

[138]

[139]

[140]

[141]
[142]
[143]

[144]

[145]

[146]

[147]
[148]
[149]

[150]

[151]

[152]
[153]
[154]

Schmidt, B.P. van Eijck, P. Verwer, D.E. Williams, A test of crystal structure
prediction of small organic molecules, Acta Crystallogr. Sect. B 56 (2000)
697714.
G.M. Day, W.D.S. Motherwell, H.L. Ammon, S.X.M. Boerrigter, R.G. Della Valle,
E. Venuti, A. Dzyabchenko, J.D. Dunitz, B. Schweizer, B.P. van Eijck, P. Erk, J.C.
Facelli, V.E. Bazterra, M.B. Ferraro, D.W.M. Hofmann, F.J.J. Leusen, C. Liang,
C.C. Pantelides, P.G. Karamertzanis, S.L. Price, T.C. Lewis, H. Nowell, A. Torrisi,
H.A. Scheraga, Y.A. Arnautova, M.U. Schmidt, P. Verwer, A third blind test of
crystal structure prediction, Acta Crystallogr. Sect. B 61 (2005) 511527.
P.H. Young, H.Y. Ando, Analysis of known crystals to design polymorph
prediction strategies, J. Pharm. Sci. 96 (2007) 12031236.
S.L. Mayo, B.D. Olafson, W.A.I. Goddard, DREIDING: a generic force eld for
molecular simulations, J. Phys. Chem. 94 (1990) 88978909.
T.C. Lewis, D.A. Tocher, G.M. Day, S.L. Price, A computational and
experimental search for polymorphs of parabanic acid a salutary tale
leading to the crystal structure of oxo-ureido-acetic acid methyl ester,
CrystEngComm 5 (2003) 39.
G.M. Day, S.L. Price, M. Leslie, Atomistic calculations of phonon frequencies
and thermodynamic quantities for crystals of rigid organic molecules, J. Phys.
Chem. B 107 (2003) 1091910933.
M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R.
Cheeseman, J.A. Montgomery Jr., T. Vreven, K.N. Kudin, J.C. Burant, J.M.
Millam, S.S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani,
N. Rega, G.A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda,
J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X.
Li, J.E. Knox, H.P. Hratchian, J.B. Cross, C. Adamo, J. Jaramillo, R. Gomperts, R.E.
Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski, P.Y.
Ayala, K. Morokuma, G.A. Voth, P. Salvador, J.J. Dannenberg, V.G. Zakrzewski,
S. Dapprich, A.D. Daniels, M.C. Strain, O. Farkas, D.K. Malick, A.D. Rabuck, K.
Raghavachari, J.B. Foresman, J.V. Ortiz, Q. Cui, A.G. Baboul, S. Clifford, J.
Cioslowski, B.B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.L.
Martin, D.J. Fox, T. Keith, M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, M.
Challacombe, P.M.W. Gill, B. Johnson, W. Chen, M.W. Wong, C. Gonzalez, J.A.
Pople, Gaussian 03, Gaussian, Inc., Pittsburgh, PA, 2003.
D.J. Willock, S.L. Price, M. Leslie, C.R.A. Catlow, The relaxation of molecularcrystal structures using a distributed multipole electrostatic model, J.
Comput. Chem. 16 (1995) 628647.
D.E. Williams, Improved intermolecular force eld for molecules containing
H, C, N, and O atoms, with application to nucleoside and peptide crystals, J.
Comput. Chem. 22 (2001) 11541166.
D.E. Williams, Improved intermolecular force eld for crystalline
hydrocarbons containing four- or three-coordinated carbon, J. Mol. Struct.
486 (1999) 321347.
D.E. Williams, Improved intermolecular force eld for crystalline
oxohydrocarbons including OHO hydrogen bonding, J. Comput. Chem. 22
(2001) 120.
G.M. Day, J.A. Zeitler, W. Jones, T. Rades, P.F. Taday, Understanding the
inuence of polymorphism on phonon spectra: lattice dynamics calculations
and terahertz spectroscopy of carbamazepine, J. Phys. Chem. B 110 (2006)
447456.
P.G. Karamertzanis, S.L. Price, Energy minimization of crystal structures
containing exible molecules, J. Chem. Theory Comput. 2 (2006) 11841199.
H. Campsteyn, L. Dupont, O. Dideberg, Structure cristalline et molculaire de
la progestrone, C21H30O2, Acta Crystallogr. B 28 (1972) 30323042.
R.W. Lancaster, P.G. Karamertzanis, A.T. Hulme, D.A. Tocher, T.C. Lewis, S.L.
Price, The polymorphism of progesterone: stabilization of a disappearing
polymorph by co-crystallization, J. Pharm. Sci. 96 (2007) 34193431.
A.J. Florence, C.K. Leech, N. Shankland, K. Shankland, A. Johnston, Control and
prediction of packing motifs: a rare occurrence of carbamazepine in a
catemeric conguration, CrystEngComm 8 (2006) 746747.
A. Gavezzotti, Calculation of intermolecular interaction energies by direct
numerical integration over electron densities. I. Electrostatic and polarization
energies in molecular crystals, J. Phys. Chem. B 106 (2002) 41454154.
C.J. Eckhardt, A. Gavezzotti, Computer simulations and analysis of structural
and energetic features of some crystalline energetic materials, J. Phys. Chem.
B 111 (2007) 34303437.
A. Gavezzotti, Hierarchies of intermolecular potentials and forces: progress
towards a quantitative evaluation, Struct. Chem. 16 (2005) 177185.
A. Gavezzotti, Modeling hydrogen bonded crystals, J. Mol. Struct. 615 (2002)
512.
S.L. Price, From crystal structure prediction to polymorph prediction:
interpreting the crystal energy landscape, Phys. Chem. Chem. Phys. 10
(2008) 19962009.
Guidance for industry. ANDAs: pharmaceutical solid polymorphism.
Chemistry, manufacturing and controls information, U.S. Food and Drug
Administration, 2007.
S.P.F. Miller, A.S. Raw, L.X. Yu, Scientic considerations of pharmaceutical
solid polymorphism in regulatory applications, in: R. Hilker (Ed.),
Polymorphism, Wiley-VCH Verlag GmBH & Co. KGaA, Weinheim, Germany,
2006, pp. 385403.
A. Llinas, J.M. Goodman, Polymorph control: past, present and future, Drug
Discov. Today 13 (2008) 198210.
J. Lucas, P. Burgess, When form equals substance: the value of form screening
in product life-cycle management, Pharma Voice (2004) 5457.
B.J. Price, J.W. Clitherow, J. Bradshaw, Aminoalkyl furan derivatives, U.S.
Patent 4128658, 1978.

J. Aaltonen et al. / European Journal of Pharmaceutics and Biopharmaceutics 71 (2009) 2337

[155] D.L. Crookes. Aminoalkyl furan derivative, U.S. patent 4521431, 1985.
[156] Electronic Orange Book: Approved Drug Products with Therapeutic
Equivalence Evaluations, U.S. Food and Drug Administration, <http://
www.fda.gov/cder/ob/default.htm>, accessed on 1st February, 2008.
[157] Applications for FDA approval to market a new drug: patent submission and
listing requirements and application of 30-month stays on approval of
abbreviated new drug applications certifying that a patent claiming a drug is
invalid or will not be infringed; Final rule, Fed. Regist. 68 (2003) 3667536712.
[158] H. Slowik, The battle for IP, In Vivo 21 (2003) 7584.
[159] M. Stuart, Transform, J&J, topiramate: new life, old products, In Vivo 22
(2004) 7172.

37

[160] L. Howard, Use of patents in drug lifecycle management, J. Generic Med. 4

(2007) 230236.
[161] D.A. Snider, W. Addicks, W. Owens, Polymorphism in generic drug product
development, Adv. Drug Del. Rev. 56 (2004) 391395.
[162] A.S. Raw, M.S. Furness, D.S. Gill, R.C. Adams, F.O. Holcombe Jr., L.X. Yu,
Regulatory considerations of pharmaceutical solid polymorphism in
Abbreviated New Drug Applications (ANDAs), Adv. Drug Del. Rev. 56 (2004)
397414.
[163] 180-Day generic drug exclusivity under the HatchWaxman amendments to
the federal food, drug, and cosmetic act, U.S. Food and Drug Administration,
1998.