Characterization of lipid nanoparticles by differential ...prescientdrugdelivery.com/wp-content/uploads/2012/... · X-ray interferences (Debye–Waller factor) and corrections specific

ews 59 (2007) 379–402www.elsevier.com/locate/addr

Advanced Drug Delivery Revi

Characterization of lipid nanoparticles by differential scanning calorimetry,X-ray and neutron scattering☆

Heike Bunjes a,⁎, Tobias Unruh b

a Institut für Pharmazeutische Technologie, Technische Universität Braunschweig, Mendelssohnstr. 1, 38106, Braunschweig, Germanyb Forschungsneutronenquelle Heinz Maier-Leibnitz, Technische Universität München, Lichtenbergstr. 1, 85747 Garching, Germany

Received 9 May 2006; accepted 24 April 2007Available online 17 May 2007

Abstract

Differential scanning calorimetry and X-ray diffraction play a prominent role in the characterization of lipid nanoparticle (LNP) dispersions.This review shortly outlines the measurement principles of these two techniques and summarizes their applications in the field of nanodispersionsof solid lipids. These methods are particularly useful for the characterization of the matrix state, polymorphism and phase behavior of thenanoparticles which may be affected by, for example, the small particle size and the composition of the dispersions. The basics of small angle X-ray and neutron scattering which are also very promising methods for the characterization of LNPs are explained in some more detail. Examplesfor their use in the area of solid LNPs regarding the evaluation of particle size effects and the formation of superstructures in the nanoparticledispersions are given. Some technical questions concerning the use of the different characterization techniques in the field of LNP research arealso addressed.© 2007 Elsevier B.V. All rights reserved.

Keywords: Colloidal drug carriers; Glycerides; Cholesterol esters; Fatty acids; Physicochemical characterization

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3802. X-ray and neutron scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380

2.1. Some considerations on the methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3802.2. Aspects of basic scattering theory important for scattering experiments on LNPs . . . . . . . . . . . . . . . . . . . . . . 381

2.2.1. Scattering of a single electron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3812.2.2. Scattering of an ensemble of electrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3822.2.3. Scattering of a volume element with homogeneous electron density — scattering from the dispersion medium . . 3822.2.4. Scattering of a dispersion of lipid nanocrystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 382

2.3. What can be learned from the complex scattering functions of LNP dispersions? . . . . . . . . . . . . . . . . . . . . . . 3842.3.1. Determination of particle size distributions on a molecular scale . . . . . . . . . . . . . . . . . . . . . . . . . . 3842.3.2. How to interpret shifts of SAXS peaks in LNP dispersions? . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3842.3.3. Scattering of medium and highly concentrated LNP dispersions . . . . . . . . . . . . . . . . . . . . . . . . . . . 386

2.4. Wide angle X-ray diffraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3862.5. Some instrumental aspects of small angle scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386

3. Differential scanning calorimetry (DSC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3884. Applications of DSC and X-ray diffraction for the characterization of LNP dispersions . . . . . . . . . . . . . . . . . . . . . . . 390

☆ This review is part of the Advanced Drug Delivery Reviews theme issue on “Lipid Nanoparticles: Recent Advances”.⁎ Corresponding author. Tel.: +49 531 3915657; fax: +49 531 3918108.E-mail address: [email protected] (H. Bunjes).

0169-409X/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.addr.2007.04.013

mailto:[email protected]

http://dx.doi.org/10.1016/j.addr.2007.04.013

380 H. Bunjes, T. Unruh / Advanced Drug Delivery Reviews 59 (2007) 379–402

4.1. Some considerations on sample preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3904.2. Observation of general thermal phenomena in LNP dispersions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3904.3. The effect of the small particle size on the thermal behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3924.4. Polymorphism and aging phenomena . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3934.5. Influences of dispersion composition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396

4.5.1. Dependence of the particle properties on the type of emulsifier . . . . . . . . . . . . . . . . . . . . . . . . . . . 3964.5.2. Use of lipid mixtures as matrix materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3974.5.3. Investigations on the effect of drug loading. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 398

4.6. Solid LNPs in semisolid formulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3995. Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399

1. Introduction

As for all drug delivery systems detailed characterization is amajor part of the research and development work on lipid nano-particle (LNP) dispersions to ensure the generation of systemswith the desired properties. Among the multitude of analyticaltechniques employed for that purpose, differential scanning calo-rimetry (DSC) and X-ray diffraction (XRD) play a prominentrole because they are able to provide structural information onthe dispersed particles. Moreover, the use of these two techniquesoften leads to complementary information on the systems ofinterest. The background of these methods will be outlined in thisreview in a relatively brief way since DSC and XRD are well-known standard techniques in the area of pharmaceutics andsince data evaluation from these methods is usually straightfor-ward. In addition to XRD, the related techniques of small angleX-ray and neutron scattering (SAXS and SANS, respectively)can give very interesting additional information on the structureof the systems. Because these techniques are less frequentlyemployed for the study of lipid nanoparticle dispersions, amore detailed introduction into their theoretical basics and theirmeasurement principles will be given with special regard to theirapplication in lipid nanoparticle research. In particular, we willconsider the specific technical challenges arising from the inves-tigation of LNP dispersions with these methods.

The second part of the review focuses on the characterization ofspecific properties of lipid nanoparticles that can be studied withDSC andXRDand also demonstrates the advantages of combiningthe two techniques. Moreover, we would like to focus the attentionof the reader on some peculiarities of the use of DSC and XRD onLNP dispersions as the investigation of nanoparticulate material inliquid dispersion is not a common part of the technical literature.

This article will only cover investigations on nanodispersionsbased on solid lipids (in some cases with the admixture of liquidones) which may, however, form solid, liquid as well as liquidcrystalline particles in the dispersed state. Moreover, to limit thescope of our contribution, we will only consider matrix-typeLNPs (e.g., emulsion droplets, suspension particles) although theanalytical techniques under discussion here have importantapplications for the characterization also of membrane-typelipid particles (e.g., liposomes, dispersions of bicontinuous cubicphases) [1–5].

It should be emphasized that nanoparticulate lipids are notonly of high interest as drug carrier systems but also play animportant role in other fields of science. As a prominentexample, dispersions of glycerides, in particular of triglycerides,are important research subjects in food technology (e.g., withrespect to the behavior of milk and cream). Therefore, muchbasic work on the behavior of such systems originates from thatarea, providing a valuable resource on basic questions in thisfield. Often DSC and XRD are employed as major analyticaltechniques in these studies [6–10]. Another example isinvestigations on biological systems such as lipoproteins forwhich also a wealth of information has been collected by thetechniques presented here [11–14].

2. X-ray and neutron scattering

2.1. Some considerations on the methods

The properties of formulations of LNPs are determinedmainlyby the manifold structures formed by the different components ofthe dispersions (lipid, drug, stabilizers, dispersion medium) andtheir interactions. Characteristic lengths d of these structuresrange from sub-atomic distances (d b 1 Å) important for e.g.crystal structure determination up to the μm range correspondingto the size of large particles or particle assemblies. The whole sizerange can be investigated by means of X-ray and neutronscattering. In general those methods detect electron (X-rayscattering) and nucleus (neutron scattering) density fluctuations,respectively, on a length scale d according to Bragg's law

2d sinh ¼ k: ð1Þ

Considering the experimentally available wavelength range0.5 Å b λ b 20 Å for neutrons and 0.5 Å b λ b 2.5 Å for X-raysthe scattering angles 2θ corresponding to the size range ofinterest are between 0.01° and 180°. Several other properties ofX-rays and neutrons such as their non-destructive nature withrespect to the sample structure, their high penetration capabilityfor organic systems allowing bulk property determination, theirsensitivity to small structural changes and the variety of theirapplications which will be highlighted in the following give thereason for the use of these probes in LNP research.

1 E=hc /λ for X-rays, E=mnh2 / (2λ2) for neutrons with h=2πħ: Planck's

constant, c: velocity of light in vacuum, mn: rest mass of neutron.

Fig. 1. Schematic representation of a general scattering experiment. Yki:f :trepresent the wave vectors of the incident, scattered (final) and transmittedbeam, respectively, with jYki;f ;t j ¼ 2p

ki;f ;t�YQ ¼ Ykf −Yki is called the scattering

vector.

381H. Bunjes, T. Unruh / Advanced Drug Delivery Reviews 59 (2007) 379–402

Corresponding to experimental equipment two differenttechniques for elastic scattering are to be distinguished:

(1) X-ray (SAXS) and neutron (SANS) small angle scatter-ing and

(2) X-ray (XRD) and neutron (ND) diffraction, also calledwide angle scattering.

The only difference between both method types is the rangeof scattering angles 2θ. While standard diffractometers coverangles between about 5° and 180°, the range between 0.01° and3° is typical for small angle instruments.

Most popular applications are the identification of crystalstructures, modifications, particle sizes and shapes as well asquantitative phase analysis (quantitative determination of thecrystalline components in a formulation) and determination ofcrystallinity indices. The interpretation of diffraction data is oftenstraightforward, e.g. for the determination of the modification oftriglyceride nanocrystals. In contrast, it may be very complicatede.g. to get reliable information on a particle size distribution ofhighly concentrated nanodispersions from SAXS or SANS data.For many investigations with X-ray and neutron scattering there ismore information about the sample in the data than usuallyextracted by the scientist. On the other hand some artifacts due to aspecial instrumentation can lead to misinterpretation as e.g.seemingly vanishing Bragg intensities for strongly texturedsamples or arbitrary multiple intensity peaks for poor powdersamples in a Kratky camera (cf. Section 2.5). In order to make itpossible for the reader to comprehend the data analysis procedures,some aspects of scattering theory and instrumental details relevantfor LNP research, concerning in particular small angle scattering,will briefly be discussed in the next sections. Some results fromscattering experiments in this field are included. For a detailedtheoretical overview the reader is referred to textbooks on X-ray[15–18] and neutron scattering [19]. A comprehensive introduc-tion to crystal geometry can be found in [20]. Aspects of neutronscattering in pharmaceutical sciences are reviewed in [21].

In contrast to the elastic case, inelastic scattering ischaracterized by an exchange of both energy and momentumbetween probe and sample [22]. This case is beyond the focus ofthis article but it should be mentioned that inelastic neutronscattering (INS) as well as quasi-elastic neutron scattering(QENS) allow to evaluate dynamic processes in LNPs on amolecular scale such as drug diffusion or internal molecularmotions. Furthermore neutron spectroscopy reveals the corre-lation of relaxation times with the spatial extension of thecorresponding process which is essential for the understandingof mass transport mechanisms. Photon correlation spectroscopy(PCS) and Raman spectroscopy, which are used for LNPinvestigations [23,24], are examples for quasi-elastic andinelastic scattering of visible light, respectively.

2.2. Aspects of basic scattering theory important for scatteringexperiments on LNPs

The validity of Bragg's law (Eq. (1)) is limited to the case ofelastic scattering which means that the X-ray or neutron does

not change its energy E during the scattering process1: jYkij ¼jYkf j (cf. Fig. 1). For this case the general equation (derived fromFig. 1 using cosine theorem) Q2 ¼ Yk

2i þYk

2f −2

YkiYkf þ4YkiYkf sin 2h reduces to

Q ¼ 2pd

¼ 2ps ¼ 4p sin hk

ð2Þ

with Q ¼ jYQ j and s=1 /d. s is often used in the literature onsmall angle scattering instead of Q. It should also be mentionedhere that from E=mc2 =hc /λ and p=mc=h /λ it is straightfor-ward that Yp ¼ ħYQ represents the momentum transferred fromthe sample to the X-ray. The same term holds for material wavesas neutrons according to predictions of de Broglie [25].

By means of some general considerations about X-rayscattering by electrons some basic results from scattering theorywhich are also valid for neutron scattering and relevant forinvestigations of LNPs will be derived phenomenologically inthe next sections. Some special effects observed for LNPs areincluded. A general expression for the calculation of thescattering function of nanocrystalline dispersions will bederived and discussed.

2.2.1. Scattering of a single electronThe kinematic scattering theory [15], which assumes that the

incident and scattered waves are not influenced by the sample,is an adequate approximation for weakly scattering samples, asnormally used in experiments with LNPs. Thus extinction,refraction and absorption are not considered in the following butcan be introduced separately [15]. The same holds for theinfluence of thermal motions of the atoms in the sample on theX-ray interferences (Debye–Waller factor) and correctionsspecific for special experimental setups such as polarizationand Lorentz factors. We consider an incident X-ray wave (cf.Fig. 1) of the form YEz

i ¼ YE0ei xt−YkiYr½ �. The wave propagates in

x direction and is polarized in z direction as indicated by thesuperscript z. For this geometry the electric field vector YE z 0

f 0 atpoint YR (cf. Fig. 2) of a wave scattered from a single electron

Fig. 2. Illustration for the calculation of the retardation of a wave scattered atYrewith respect to the point of origin; R ¼ jYRj, Ys ¼ R

Y

R , Ys0 ¼kYij kYi j

, X1 ¼ Yre �Ys0,X2 ¼ jYR−Yrej¼ RY − rYeð Þ�Ys

cos a .


positioned in the coordinate system's point of origin resultsfrom Maxwell's theory and is given by

YE z 0f 0 ¼

e2 sin hzmec2R

EY0 ei xt−Ykf Yr½ � ð3Þ

with e: elementary charge and θz: angle between z and x′. Thiswave propagates along x′ and is polarized in z′ direction. Thescattering amplitude YE z0

freof an electron positioned at Yre is

according to Fig. 2 and using Fraunhofer's approximation X1 þX2 ¼aY0

RþYre Ys0−Ysð Þ given by

YE z0fre ¼

e2 sinhzmec2R

YE0ei xt−2p

k RþYreYkf −Ykið Þ½ �: ð4Þ

The intensity scattered from a single electron can becalculated by multiplication of YE z0

frewith its complex conjugate:

Ie ¼ e4 sin 2hzm2

ec4R2

jYE 20 j ð5Þ

2.2.2. Scattering of an ensemble of electronsThe intensity scattered from n electrons can accordingly be

calculated after adding up Eq. (4) for all n electrons:

InYQ� �

¼ IeYQ� �

�Xnk¼1

Xnj¼1

cos Yre; j−Yre;k� � �YQ� �

ð6Þ

An important interpretation of this equation is that the totalscattered intensity of an ensemble of n electrons is proportional ton2 if the electrons are within a volume with dimensions smallcompared to the X-ray's wavelength and therefore Yre; j−Yre;k

� ��YQ≪ 18j; k. If Yre; j−Yre;k

� � �YQ≫ 18j; k, which corresponds towidely spread electrons, the total scattered intensity is propor-tional only to n. This is because for j≠k the argument of thecosine is equally distributed over the whole codomain during thesummation leading to a mean contribution to the sum of zero. Theremaining terms for j=k lead to the mentioned proportionality.

From these considerations we can now understand the resultfrom small angle scattering theory that for widely separatedparticles the scattering intensity is proportional to the square of theelectron density of a single particle but only proportional to thenumber of particles [18]. The following consequence for the totalscattering of an LNP dispersion might be surprising at a first view:We compare the total scattering intensities of two low concentratedLNP dispersions both with the same total mass of particles thatmeans with the same total number of electrons inside the particles.The first dispersion has half the number of particles but with doubleweight of each compared to the second dispersion. Whereas the

total mass of the scattering particles is the same for both dispersionsthe total scattering of the first dispersion has twice the intensity ofthe one of the second dispersion.

2.2.3. Scattering of a volume element with homogeneouselectron density — scattering from the dispersion medium

Assuming a large sample with a homogeneous electrondensity the same argument introduced above is valid withregard to the calculation of the scattered intensity to a fixed YQ.For small distances ljk ¼ jYre;j−Yre;k j of infinitesimal volumeelements with the constant electron density ρe all cosine valuesare positive. But for increasing ljk the cosine values will changealternately their signs. In this case the total scattered intensitywill be essentially zero except for extremely small angles. If thesample is large enough these angles are experimentally notaccessible. According to Babinet's theorem it can be followedthat the scattering intensity I YQ

� �of a small particle with

homogeneous ρe cannot be distinguished from the scattering ofan infinite large sample with the same electron density but witha cut-out of the size of the small particle. By means of thistheorem the amplitude YE0 � SkD of the X-ray scattered by thedispersion medium surrounding a crystal k of parallelepipedicshape (cf. Section 2.2.4) can be calculated:

−SkD ¼ VkqDNk1N

k2N

k3

j3

j¼1

1

i Ykf −Yki

� �Yaj ei Ykf −Ykið ÞYajNk

j

h i−1

!ð7Þ

(Yaj: fundamental lattice vectors of crystal k, Njk: number of unit

cells of crystal k inYaj direction, Vk: total volume of crystal, ρD:mean electron density of dispersion medium).

Another important conclusion of the discussion above is that thescattering of a particle with an electron density ρe cannot beobserved in a dispersion medium with the same electron density.This holds only for homogeneous electron densities. In realparticles or dispersionmedia the electron densities vary in space ona certain length scale l. But if we talk about Q-values small withrespect to 1/l these fluctuations cannot be observed. Therefore thescattering intensity of e.g. an LNP dispersion is proportional to thedifference of the averaged electron densities of the particles and thedispersion medium (called the scattering contrast). This iscorrespondingly true for neutron scattering, where the scatteringcontrast is easier to change. This holds especially for aqueoussuspensions where water can be partly or fully exchanged by D2O.The coherent scattering cross section of deuterium is more thanthree times larger than that of hydrogen [26]. Hence, for typicalLNP dispersions the scattering contrast between the particles andthe dispersion medium is for neutrons (when using D2O) muchlarger compared to that of X-rays. This compensates the muchlower primary beam intensities at neutron instruments compared toX-ray beam lines to a large extent and measuring times of onlysomeminutes per diffraction pattern are required for good statistics.

2.2.4. Scattering of a dispersion of lipid nanocrystalsCrystalline nanoparticles of triglycerides in their stable

β-modification exhibit the shape of a parallelepiped with onedimension much smaller compared to the others (Fig. 17 bottom).

Fig. 3. Visualization of the position vectors Yre ¼ YRV0 þYRk þ m1Ya1þ

m2Ya2 þ m3

Ya3 þYrj þY~R, which refer to an infinitesimal volume element(integration variable Y~R) of the jth atom in the unit cell designated by the triple(m1, m2, m3) of the kth nanocrystal in the dispersion. Each YRk defines the point oforigin of the coordinate system of a particular particle k. Each particle(parallelepiped) consists of m1, m2, and m3 unit cells in x′, y′ and z′ direction,respectively, of its coordinate system. The orientation of this coordinate system hasto be defined by transformationmatrices with respect to the laboratory system x, y, z.


The alkyl chains of the triglyceride molecules (long c-axis of theunit cell) are oriented along that small dimension, which isreferred to as thickness of the crystal. A simplified (2D)schematic representation of such a nanocrystal (with a thicknessequal to that of two unit cell heights) is displayed in Fig. 3. On thebasis of crystals of parallelepipedic shape it is also possible toapproximate the shape of other crystals. Therefore this modelseems to be well suited to be used for simulation calculations ofX-ray patterns of nanocrystalline dispersions and will bedescribed in this section.

The amplitude of the electric field of the scattered wave froma volume element of a dispersion of nanocrystals, which is smallcompared to the coherence volume of the X-rays,2 can becalculated using the vectors Yre defined in Fig. 3 by

YE z0tot

YQ� �

¼XNk

k¼1

XNk1−1

m1¼0

XNk2−1

m2¼0

XNk3−1

m3¼0

XNA

j¼1

ZVjA

YE z0freq

jedV : ð8Þ

The scattering contribution of each of the NA atoms in theunit cell is calculated by the integration of the amplitudes YE z0

frescattered by the electron density ρe

j. The integration variable isthe vector Y~R (cf. Fig. 3) and the integration is performed overvolume VA

j of the jth atom in the unit cell. This integration doesnot need to be performed for simulation calculations but theresults are tabulated as atomic form factors in the InternationalTables for X-ray Crystallography [28]. After the calculation ofthe scattering contribution of all atoms of a single unit cell wehave to sum up the contributions of all unit cells in the crystaldesignated by the vectors mjYaj with mj∈ [0,Nj

k−1]. The last

2 The coherence volume designated a volume element of the sample which isilluminated by the primary beam and all X-rays or neutrons scattered from thatvolume can at least partially interfere with each other. The size of that volumedepends only on the instrumental parameters as e.g. monochromaticity of theprimary beam, size of the source, apertures before and behind the sample. It canbe estimated from the volume of a cuboid of length ξl, width ξy and heightξz ·ξl is parallel to the propagation direction of the wave and therefore called thelongitudinal coherence length. ξy,z is called the transverse coherence length in yand z directions, respectively. For more details cf. footnote 3 and [27].

sum adds the scattering contributions of all Nk nanocrystals inthe respected sample volume referenced by YRk.

It can be stated here that the coherence volume for X-ray aswell as for neutron diffraction experiments exceeds the size ofparticles and particle associates of interest in LNP research,which means that the diffraction patterns should include allrelevant structural information.3

For a dispersion of crystals of parallelepipedic shapeYE z0

totYQ� �

can be written as:

YE zVtot

YQ� �

¼ YE0

f �XNk

k¼1

eiYRk �YQ � Fk � Gk−SkD

� �n oð9Þ

with the structure amplitude Fk comprising the scattering of theatoms in the unit cell of the kth crystal, the lattice amplitude Gk

describing the interference of the periodically arranged unit cellsand SD

k being an approximated term which summarizes thescattering of a quasi continuous dispersion medium. It is obviousthat Fk, Gk and SD

k are functions of YQ. Fk and SDk vary smoothly

with YQ, whereas Gk exhibits distinct sharp maxima around YQ ¼YG which are responsible for the Bragg reflections in I YQ

� �.YG

represents any linear combination of the fundamental vectors ofthe reciprocal lattice [20].

The product of YE z0tot with its complex conjugate gives the

intensity of the scattered X-rays:

I YQ� �

−YE z0tot

YQ� �

� YE z0tot

YQ� �� 4

: ð10Þ

From this multiplication cross-terms arise besides the |FkGk|2

(|Fk|2 and |Gk|2 are called structure factor and lattice factor,respectively) and |SD

k |2 terms. For a single particle there is onlyone cross-term −2Re(SDk FkGk) (Re(x) designates the real part ofx). This cross-term is some kind of interference term betweenparticle and solvent scattering, which influences the small anglescattering of LNP dispersions significantly as demonstrated inSection 2.3.2. For an ensemble of particles also interparticleterms arise, which describe the interference of the scattering ofdifferent particles. These terms influence and dominate in somecases the small angle scattering curves of medium and highlyconcentrated SLN dispersions as discussed in Section 2.3.3.

According to the limited size of the coherence volume thevector YRV0 (cf. Fig. 3) shall designate the point of origin of acoherence volume element. A crude approximation to calculatethe scattering of the whole sample is to multiply I YQ

� �with the

total sample volume divided by the coherence volume.

3 The transverse (spatial) coherence lengths of X-rays and neutrons withwavelength λ0 in diffraction experiments can be estimated by ξx,y ≈ λ0lc / ly,z(lc: distance from source to sample (collimation length), ly,z: lateral dimensionsof the source in y and z direction, respectively). If the primary beam iscollimated by apertures in a way that the full source size is not seen by thesample the lateral aperture dimensions and the distance from aperture to samplehave to replace ly,z and lc, respectively. The longitudinal (temporal) coherencelength ξ1 can be estimated according to the approximated equation known fromclassical optics l ¼ k20

Dk and the retardation l=2dsinh for Bragg diffraction byl ¼ k0

Dk2kQ , where Δλ is the full width at half maximum (FWHM) of the

wavelength spectrum of the primary beam.


2.3. What can be learned from the complex scattering functionsof LNP dispersions?

2.3.1. Determination of particle size distributions on amolecular scale

For small crystals of only a few unit cells thickness the Braggpeaks are significantly broadened due to the lattice amplitude,which is not yet sharply peaked. Therefore information on theparticle thickness distribution of an LNP dispersion is containedin the width and shape of the Bragg reflections. From a singleBragg reflection at YQ ¼ YG the thickness distribution of theparticles parallel to G can be deduced. The analysis of differentreflections can be used to distinguish between particle thick-nesses along the corresponding directions. If one or more higherorders of a reflection can be measured simultaneously strain andsize effects which both contribute to Bragg peak broadening canbe separated. For the case of triglyceride particles also a singleline evaluation excludes the influence of strain effects [29].

Nanoparticles of β-triglycerides are platelet-like shaped witha diameter typically in the range between 50 nm and 300 nm anda thickness about a factor ten smaller. The size and aspect ratioof the particles are influenced by the emulsifiers, the type andmodification of the triglyceride and the preparation conditions[30–32]. The large surfaces of the platelets are parallel to thecrystallographic (001)-plane. Thus the thickness distribution ofthe platelets can be evaluated from the (001)-reflection whichoccurs in the small angle regime. It is obvious that a single

Fig. 4. Difference diffraction patterns of a temperature resolved SAXSmeasurement of a tripalmitin nanosuspension. During the experiment thetemperature of the sample was increased stepwise. The displayed curves resultfrom subtraction of diffraction patterns recorded at temperatures listed in thelegend. The calculated average thickness of the particles which melted in thecorresponding temperature ranges is labeled for each curve. The curves areshifted along the y-axis for clarity. It is clearly visible that the position of the(001)-reflection is shifted to higher s-values with increasing particle thickness.For 50 nm thick particles the value for polycrystalline tripalmitin, marked by thedashed line, is reached. The increasing width of the peak with decreasing particlethickness is also visible. The sample was prepared by melt homogenization of10% tripalmitin in water with a high pressure homogenizer using 3.2%phospholipid and 0.8% sodium glycocholate as stabilizers. After recrystalliza-tion of the fat at room temperature the sample was diluted with pure water to atotal tripalmitin concentration of 3% to reduce particle interactions. The PCSparticle size was about 140 nm (z-average). The measurement was performed atthe ID02 beam line of the ESRF, Grenoble.

diffraction peak does not contain enough information in order toextract a complex size distribution function. In the case oftriglycerides it is, however, possible to collect many SAXSpatterns over the broad melting temperature range of the particles(cf. Section 4.3). The diffractograms produced by subtraction oftwo SAXS patterns collected at only slightly different tempera-tures yield information on the particles melted in that temperaturerange. Such difference diffraction patterns but with quite largetemperature steps are displayed in Fig. 4. It could be dem-onstrated that the melting temperature of triglyceride nanopar-ticles decreases with their decreasing thickness, and particle sizedistributions with molecular resolution could be determined[29,33].

2.3.2. How to interpret shifts of SAXS peaks in LNPdispersions?

As discussed in Section 2.2.4 the calculation of I YQ� �

evenwith the simplified term for YE z0

totYQ� �

of Eq. (9) gives a largenumber of cross-terms of reasonable intensity contributions.This is because for crystalline nanoparticles the size of their unitcell, the crystal's dimensions and the interparticle distances in adispersion are within the same order of magnitude. This alsoholds for liquid crystalline and to some extent even for non-crystalline particles.

The relevance of the considerations above shall be demon-strated again for nanocrystalline triglyceride dispersions. From thetemperature resolved SAXS experiments discussed above theposition of the Bragg reflection on the s-scale for each particle sizefraction can also be determined. It can be seen from Fig. 4 that thisposition is shifted continuously to lower s-values for decreasingparticle thicknesses. Using Bragg's law the peak shift couldmisleadingly be interpreted as an increase of the (001)-interplanarspacing with decreasing thicknesses of the nanocrystals. Thisinterpretation would be in contradiction to the observations thatlattice constants of nanocrystals decrease with decreasing particlesize due to the influence of the interfacial tension.

Therefore simulation calculations were performed on thebasis of Eq. (9). In Fig. 5 calculated scattering curves oftripalmitin nanoparticles with thicknesses of 1, 2 and moremolecular layers are displayed. The position of the (001)-Braggreflection of polycrystalline tripalmitin (powder sample) ismarked in each subfigure by a vertical line. For the calculationsthe crystallographic data of a large tripalmitin crystal were used.Although the (001)-interplanar distance was kept constant in thecalculations the position of the calculated (001)-Bragg reflec-tion is continuously shifted to lower s-values with decreasingparticle thickness. The simulation results are quantitatively ingood agreement with the experimental results as illustrated inFig. 6 or in [33].

The reason for the peak shift is an interference of the latticeamplitude, which represents still a broad peak function on the s-scale due to the small number of unit cells along YG 001ð Þ, thestructure amplitude and the scattering amplitude from thedispersion medium including the stabilizer layer. It isremarkable that the simulation results cannot be reproducedwithout the phospholipid layer. In this case the peak shift isnegligible and not consistent with the SAXS data. In contrast to

Fig. 5. SAXS patterns calculated for single tripalmitin nanoparticles surrounded by a phospholipid monolayer in an aqueous solution for different particle thicknesses(number of unit cells parallel to crystallographic c-axis). The dimensions of the crystals parallel to a- and b-axes were 120 nm and 100 nm respectively. The structurefactor was calculated from the crystal data given in [34]. The phospholipid layer was approximated by two layers around the crystal of homogeneous electron densityeach with ρ=310 electrons per nm3 (hydrophobic tail part, thickness l=0.8 nm) and ρ=350 electrons per nm3 (hydrophilic headgroup part, l=1.3 nm), respectively(approximated from DPPC double layer simulation results e.g. [35]). The scattering amplitudes of these layers and the aqueous solution (ρ=333 electrons per nm3)were calculated according to Eq. (7). When using approximated electron densities from DPPC double layer simulation results (e.g. ρ=228 electrons per nm3

(hydrophobic tail part, thickness l=1.1 nm) and ρ=398 electrons per nm3 (hydrophilic headgroup part, l=1.2 nm) [35]) the calculated SAXS patterns do not fit to theexperimental data. Displayed is the scattering intensity (for clarity in different arbitrary units), averaged over all particle orientations, as calculated by numericalintegration. The dashed line in the second subfigure represents the scattering of the same particle (including the phospholipid layer) used for the solid curve on the sameintensity scale but excluding the scattering of the dispersion medium (particle in vacuum).


recent cryo transmission electron investigations [36] thephospholipid layer could now clearly be distinguished fromthe triglyceride by X-ray scattering. Furthermore it was foundthat the geometry as well as the electron density distribution ofthe phospholipid layer deviates significantly from the situationin phospholipid double layers [37].

The shift of the (001)-peak is strongly influenced by theelectron density difference between the continuous phase and thenanoparticle. As illustrated by the dashed line in the secondsubfigure of Fig. 5 the small angle scattering of a particle invacuum is considerablymore intense compared to that of the same

Fig. 6. Dependence of the (001)-spacing d001 on the particle thickness D. Thedata (circles) have been extracted from SAXS patterns of a tripalmitinnanodispersion characterized in the caption of Fig. 4. The data analysis hasbeen performed as described in [33]. The triangles correspond to the valuesdeduced from the simulations presented in Fig. 5.

particle in a dispersion medium, due to the increased electrondensity contrast, and the shapes of the two scattering curves arecompletely different. In the case of tripalmitin the simulationspredict a shift to higher s-values when the electron density of thedispersion medium exceeds the one of water. Such an investiga-tion is hard to do for X-rays. But an analogous SANS experimentwith deuterated fat should be possible.

Fig. 7. SAXS patterns of a native (10 wt.%) and diluted tripalmitinnanosuspensions recorded at the X33 beam line of the EMBL at DESY,Hamburg, Germany. The curves are shifted along the intensity scale for bettervisualization. The features due to the formation of stacked lamellae are markedwith arrows for the 10% curve. With increasing Q-values the intensity of thesefeatures decreases but in the regime of the broad tripalmitin (001)-reflectionaround s=0.248 nm−1 they are again amplified by interference with the Braggscattering. The repeat distance of the lamellae is about 33 nm for the 10%dispersion and 42 nm for the 7% dispersion. For the dispersions with only 3 wt.%of tripalmitin no stacked lamellae are observed.


With increasing particle thickness the (001)-reflection nar-rows due to the shape of the lattice amplitude, which becomesmore and more dominant. For 20 unit cells along the c-axis atypical Bragg reflection can be observed. The significantdifference between the scattering of a single unit cell and asingle tripalmitin layer is due to the different small anglescattering of both particles when YQ is perpendicular to YG 001ð Þ.

2.3.3. Scattering of medium and highly concentrated LNPdispersions

Eq. (9) includes the interference effects between differentnanocrystals. These effects can be avoided by dilution of theinvestigated dispersions so that the particles are not associatedas illustrated in Fig. 7. For many systems one is, however,interested in the native formulation. For triglyceride dispersionsthe formation of stacked lamellae and even liquid crystallinearrangements has been reported [38–40]. For the case ofstacked lamellae additional features in the Q-range below the(001)-reflection appear (cf. Fig. 7). A strong interference ofhigher orders of these features with the (001)-reflection leads toits partial (Fig. 7) or complete deformation as observed for DNAloaded tripalmitin nanodispersions [41]. In the latter case thestacked lamellae can be regarded as aggregates and thus thesmall angle features can also be observed at low particleconcentrations. From the additional features the interparticledistances in the stacked lamellae and their relative frequencycan be deduced. It should, however, be mentioned here that dueto the complex interference of many cross-terms the interfer-ence maxima are not equidistant [38] as expected for a lamellarphase, which complicates a detailed analysis.

2.4. Wide angle X-ray diffraction

Wide angle X-ray diffraction is well-known in LNP researchand can be regarded as an established method. Therefore it will beonly briefly introduced here. Most X-ray studies that have beenpublished for lipid nanoparticles are wide angle diffractioninvestigations. This is because much important information canbe revealed from the crystalline state of drugs and excipients of anLNP formulation. First of all the crystalline ingredients of adispersion could be analyzed quantitatively according to thecharacteristic diffraction patterns of the known structures (phaseanalysis). It should however be kept inmind, that the sensitivity ofthe method is low which means that an analysis of a component

Fig. 8. Schematic representation of a simplified pinhole small angle scattering apparatsample as e.g. l1 and l2. The detector plane is perpendicular to the primary beam di

of less than 0.5 wt.% with respect to the total sample mass isdifficult.

If a crystalline component is detected one gets alsoinformation on its crystal modification [42,43,31,44,45–47].This is particularly important with respect to the matrix lipidswhich are very often polymorphic substances. Modificationchanges after preparation of a formulation are well-known toe.g. segregate the drug out of the nanocrystals and can therebyinfluence the drug release properties of the dispersions [48].

If the intensity of the used X-ray instrument is high enougheven fast transitions of modifications can be resolved by subse-quent detection of diffraction patterns in time [30,31,43,46,47]. AnX-ray/DSC combination, which is very useful for LNP research,has been used at the beam line BL-15A of the source ‘PhotonFactory’, Tsukuba, Japan [49] and at beam line 5.2 at ELETTRAsynchrotron, Trieste, Italy [10]. Both complementary methods canbe applied simultaneously on the same sample and thus a directcorrelation of DSC signals and structural transitions is obtained.

2.5. Some instrumental aspects of small angle scattering

According to Eq. (1) it should be possible to overcome theproblem of measuring at very small angles, which is experimen-tally intricate, when using longer wavelengths λ. The absorptionof X-rays with wavelengths above 2 Å by the sample is, however,so high that the scattered beam is strongly attenuated andabsorption corrections are difficult to apply [18]. For neutrons itis possible to use wavelengths up to more than 20 Å but theintensity of the available neutron beam decreases dramaticallywith decreasing neutron energy for all available neutron sources.

The basic principles of a small angle apparatus can beexplained corresponding to the simple drawing in Fig. 8. First ofall a radiation source for X-rays and neutrons, respectively, isneeded. Conventional X-ray generators which produce X-raysby electron bombardment on a fixed or rotating anode target areused in commercially available laboratory instruments for bothSAXS and XRD. The target material determines the resultingX-ray energy which is typically in the energy range between5.4 keV (CrKα

) and 24.9 keV (AgKβ1). Besides the ‘Bremsstrah-

lung’ the spectrum of a copper target, which is used most oftenon pharmaceutical samples, is dominated by the two intenselines CuKα1

at λ=1.5405 Å and CuKα2at λ=1.5443 Å. The

second line has about half the intensity of the first one and canbe removed by a primary monochromator. If no primary

us. The detector can be moved in the detector pipe to different distances from therection and is determined by its distance to the sample.


monochromator is used the CuKβline is usually reduced in

intensity by a Ni-filter.In the case of weakly scattering samples it is convenient to

use more intense X-ray sources available at electron (orpositron) synchrotrons. The electrons are accelerated up toseveral GeV and stored in large circular tubes. The X-rays aregenerated by deflection of the electrons by a magnetic field.Because of the relativistic velocity of the electrons the generatedX-ray beam is extremely intense at an extraordinary brillianceB ¼ N=t

dXdF dk=kð Þ (N/t: number of emitted photons per time, dΩ:solid angle, dF: area of source, dλ/λ: relative bandwidth). Theuse of special arrangements of periodic fields leads to evenhigher intensity and brilliance for polychromatic (wiggler) andmonochromatic (undulator) beams. Measuring times for a smallangle diffraction pattern at modern synchrotron sources are inthe subsecond range. Beam sizes smaller than 100 × 100 μm2

are available.The measured quantity of the scattering experiment is the

intensity as a function of the scattering angle 2θ. For goodangular resolution the primary beam divergency needs to besmall and the resolution of the detector must be high. A simplemethod to reduce the divergency of the primary beam is to use acollimator made of two pinholes separated by the distance c(collimation length). Increasing c will reduce the divergencybut also the intensity. Focusing optics are often used to achievehigher intensity. The angular resolution determined by thespatial resolution of the detector (Δr) and the distance l betweenthe sample and the detector can be evaluated for small anglesaccording to Δ(2θ) = arctan(x / l)− arctan((x+Δr) / l)≈Δr / lwith x being an arbitrary distance between the positions of theprimary beam and the scattered beam in the detector plane.Increasing l thus increases angular resolution and assuming a

Fig. 9. 2D-detector images of a SAXS (left side, recorded at ID02 beam line, ESRF,pattern of tripalmitin nanodispersions. The tripalmitin concentrations were 19% fomedium for the SANS experiment was prepared using D2O instead of H2O. For both imarrows. The (001)-Bragg reflections are marked by white arrows. The strong radiacrystalline domains of particles oriented in parallel. In the X-ray image the broad shadvisible in the center of the neutron image.

fixed size of the beam stop extends the angular range to smallerscattering angles but reduces the largest detectable scatteringangle due to the finite detector size.

When a simple counter tube is used as detector it is necessaryto scan the detector plane by moving the detector step by step inthe plane. Because of the extended part of the simultaneouslydetected solid angle, reducing the measuring times significantly,it is more convenient to use one- or two-dimensional positionsensitive detectors (1D/2D detectors), such as gas filled delayline detectors, image plate detectors or CCD cameras. Suchdetectors are available for both X-rays and neutrons. Whenusing 1D detectors the information of the radial intensity dis-tribution with respect to the primary beam is lost. Furthermoreone has to pay attention to the fact that the total intensityscattered to angles 2θ±d(2θ) / 2 is distributed over the solidangle dΩ=2π sin(2θ)d(2θ) which for small scattering anglesincreases approximately linearly with increasing 2θ. The detectedintensities can be corrected when assuming a homogeneousintensity distribution over dΩ which is in general not valid forpowder samples with any texture or other interparticle ordering.Such particle arrangements can be investigated using 2Ddetectors. Detector images of an X-ray and a neutron smallangle scattering experiment, respectively, on a tripalmitin nano-dispersion are displayed in Fig. 9. The inhomogeneous radialintensity distribution which is correlated to the anisotropic parti-cle arrangement in these dispersions can be seen clearly. Whenmeasuring these samples with a 1D detector it depends on theparticular sample orientation whether the superstructure is visi-ble at all or to what extent. In any case without further informa-tion the detected intensities are hard to analyze quantitatively.

For laboratory use the very compact setup of a Kratkycamera is often used for SAXS experiments. This camera type

Grenoble) and a SANS (right side, recorded at D11 instrument, ILL, Grenoble)r the SAXS experiment and 15% for the SANS measurement. The dispersionages the first interference peaks due to the particle stacking are marked by black

l anisotropy of all interference maxima indicates the formation of large liquidow of the beam stop for the primary beam is visible. The rectangular beam stop is

Fig. 10. SAXS patterns of a liquid crystalline hexagonal phase (dobanol 40 withwater) recorded using a Kratky camera (HECUS M.Braun, Graz). The verticallines mark the positions of the expected Bragg reflections. With the standardsetup of the camera and a horizontal slit width of about 1 cm the diffractionpattern represented by the solid line was observed. The dashed line represents ameasurement with an extra horizontal collimation equipment (point focus of theX-ray tube and 0.5 mm width of the horizontal slit).


uses a horizontally oriented line shaped beam profile (of about1 cm length) which is well collimated vertically but less wellhorizontally. Because of the camera geometry it is possible torecord scattering curves down to s≈0.01 nm−1. The diffractionpatterns are however ‘smeared out’ to small Q-values becauseof the poor horizontal collimation of the beam. This effectbecomes more and more crucial for decreasing scatteringangles. A ‘desmearing’ procedure has to be applied to get thetrue scattering function. These procedures are based on theassumption that the investigated sample is a perfect powdersample, which means that the investigated particles are smalland randomly oriented over the whole sample in the beam. InFig. 10 an uncorrected SAXS pattern of the hexagonal liquidcrystalline phase of the non-ionic surfactant dobanol 40 in wateris displayed. Instead of the three expected reflections for ahexagonal phase (ratios: 1 :

ffiffiffiffiffiffiffiffiffi3 : 2

p), indicated by vertical lines,

many partly overlapping features are observed. The reason for

Fig. 11. Schematic representation of the experimental setup in heat flux D

this artificial pattern is that the liquid crystalline phase exhibits astructure with large homogeneous domains. Some domains areoriented such that they fulfill Bragg's law and because thedomains are large their scattering intensity is high. The detectedscattering angles of these domains equal the expected value orare lower if Bragg's condition is fulfilled for a horizontallydivergent beam. Thus each peak of the detected SAXS patternbelongs to one or more special liquid crystalline domains of thestick shaped sample. When including a horizontal slitcollimation option such that the primary beam is limited to0.5 mm in width the expected pattern can be observed (dashedline in Fig. 10). Now there is of course nearly no smearing butsome peak intensities have nearly vanished. The same effect hasto be expected for concentrated LNP dispersions. Therefore theauthors of this article were not able to detect e.g. stackedlamellae of triglyceride dispersions using a Kratky camera.

3. Differential scanning calorimetry (DSC)

Since DSC is a well-known technique in pharmaceuticalresearch and development only a short overview will be givenover the basic principles of DSC measurements and instrumen-tation. For detailed information, the reader is referred to thespecialized literature, e.g., several comprehensive textbooksexplaining the measurement principles and their applications[50,51]. Concerning particularly the pharmaceutical field, thebook by Ford and Timmins gives an exhaustive overview overthe relevant thermal analysis techniques and applications [52].Moreover, there are several introductory articles covering theprinciples of the measurement technique and their applicationsin the pharmaceutical area, e.g., [53,54].

Structural alterations of materials are accompanied by heatexchanges, e.g., uptake of heat during melting or emission ofheat during crystallization. DSC is designed to measure theseheat exchanges during controlled temperature programs andallows to draw conclusions on the structural properties of asample. It should be borne in mind, however, that, although

SC (top, disk-type system) and power compensation DSC (bottom).

Fig. 12. Comparison of the DSCmelting curves of trimyristin as rawmaterial and ascolloidal dispersion (10% trimyristin, 2.4% soybean lecithin, 0.6% sodiumglycocholate in an aqueous phase containing 2.25% glycerol and 0.01% thiomersal).


DSC is able to monitor and quantify even minute thermal eventsin the sample (depending on the sensitivity of the instrument, ofcourse) and to identify the temperatures at which these eventsoccur, it is a technique which does not directly reveal the causeof a thermal event. The exact nature of the thermal transitionshas to be determined with complementary methods such asmicroscopic observations, thermogravimetry, X-ray diffractionor spectroscopic techniques to distinguish, for example,between melting, polymorphic transitions, loss of water fromhydrates or decomposition of the substance. In common DSCinvestigations the respective sample is heated or cooled at acontrolled rate and the heat flow into or out of the sample ismonitored in a quantitative way. Although isothermal measure-ments are, in principle, also possible with conventional DSCequipment, they are not commonly performed, mainly sincesuch measurements usually require a very high sensitivity of theinstrument and thus specialized equipment (microcalorimetry).

There are two principal types of instruments used for DSCinvestigations, heat flux DSC and power compensation DSC(Fig. 11). Both contain two identical sample holders which areheated (or cooled), usually in a linear way.4 One sample holdercontains a crucible filled with sample, the other a referencecrucible. Under ideally symmetric conditions (perfectly sym-metric instrumental setup and identical properties of sample andreference) the temperature in both crucibles increases at thesame rate during a heating program and there is no temperaturedifference between sample and reference as long as no structuralchange occurs in the sample. When the sample undergoes astructural change, as for example a melting transition, it takes upenergy, i.e. heat, from the system. As during the course of thetransition all energy provided by heating is used for thetransition, the sample temperature does not increase any longer.This leads to a temperature difference between the sample and

4 Temperature modulated DSC, where the linear heating or cooling rate issuperimposed with a sinusoidal or other type of temperature modulation, has notyet been used for the study of LNP dispersions and will thus not be covered here.

the reference crucible which continues to follow the instrumen-tal heating program. In heat flux DSC, this temperaturedifference is measured and converted into a heat flow signalthrough a calibration procedure. In power compensation DSC,where the two sample holders (microfurnaces) are supplied withindividual heaters, the detection of any temperature differencebetween the furnaces leads to an immediate adjustment of theheating power in such a way that the temperature is kept(almost) identical in both furnaces. The additional (or reduced)heating power is directly correlated with the heat flow into orout of the sample. Adequate calibration is, however, requiredalso for this measurement technique. After the calibrationprocedure, there are, in principle, no significant differences inthe results obtained from the two different measurementtechniques in routine experiments (although the experimentalresults may depend significantly on the sensitivity andresolution of the respective instrument). By convention,endothermic events are displayed as upwards signals in powercompensation DSC and as downwards signals in heat flux DSC.

The signal obtained in DSC measurements is the heat flow asa function of time or temperature. For first-order transitions, likemelting, crystallization or polymorphic transitions, signals ofpeak-like shape are observed (Fig. 12). Integration of the signalsprovides the heat exchanged during the transition. Glasstransitions (which are, however, less important with respect tothe characterization of LNP dispersions) are characterized by astep-like change of the baseline position. As an additionalcomponent, DSC curves usually show a deviation of thebaseline from the zero position which, in addition to certaininstrumental parameters, reflects the differences in heat capacitybetween the sample and the reference, which is usually anempty crucible. While this deviation is relatively small when thesample consists of small amounts of organic material (e.g., bulkfats) it may be quite large when aqueous dispersions aremeasured due to the high sample mass required and the largeheat capacity of water (Fig. 12). In most cases, this deviation isirrelevant for the measurements as it can simply be subtracted.

Fig. 13. DSC heating curves of pure indium showing the influence of heatingrate on the peak maximum temperature (Tmax). The influence of the heating rateon the extrapolated peak onset temperature (Tonset) is much smaller. All curveswere recorded with the same temperature calibration.


In specific applications, it may, however, be useful to balancethe heat capacities between sample and reference by filling thereference crucible with an adequate amount of inert material.Some instruments, in particular those for high sensitivityinvestigations, even require this kind of heat capacity balancing.

For pure substances, the temperature of a first-order transitioncan be derived in a straightforward way as the extrapolated onsettemperature of the respective thermal event of the DSC curve(Fig. 13). The extrapolated onset temperature does only littledepend on the experimental conditions and influences can becorrected for by calibration. In contrast, other characteristicpoints of the transition, in particular the peak maximumtemperature, highly depend on experimental parameters: Dueto the thermal lag of the instrument, the use of higher heatingrates will shift the maximum of the signal to higher temperatures(Fig. 13). For the same reason, the peak maximum temperaturealso tends to shift slightly to higher temperature with increasingsample mass. As outlined in more detail below (cf. Section 4.3),the determination of representative melting temperatures may bemuch more problematic for dispersed materials, in particularwhen the particle sizes are in the nanometer range. For suchsamples, the shape of the resulting transition frequently deviatessignificantly from that of a “neat” signal of a pure bulksubstance, often making the determination of a meaningful onsettemperature impossible. The shapes of signals that correspond tothe crystallization of the dispersed particles upon cooling areusually less complex and an extrapolated onset temperature can,in most cases, be derived without problems. For exothermictransitions the onset temperature is a much more realistic valuethan the peak maximum temperature since the sampletemperature during the transition is not very well controlleddue to the evolution of heat as a result of crystallization [55,56].

With modern DSC equipment the thermal behavior ofsamples can be monitored between subambient temperatures(depending on the cooling device) and several hundred degreescentigrade. Nitrogen or helium is used as purge gases to avoidreactions with the atmosphere. For measurements in the lipidnanoparticle field, the temperature range between about −60 °C(for freeze-drying applications) and ∼350 °C (highest meltingpoints of organic substances) is the most interesting, withparticular regard to the temperature range of liquid water.Typical scan rates are between 1 and 10 °C/min although, inparticular, lower rates may be used for specific applications or inspecialized equipment. While low scan rates are preferable interms of peak resolution and investigation of the sample close toequilibrium, high scan rates increase the sensitivity of themeasurement as they lead to the exchange of heat within acomparatively short time period. It should be borne in mind thatthe scan rate may also influence the course of temperature-related processes within the sample. For example, high scanrates can suppress gradual transformation of metastabletriglyceride modifications upon heating (cf. Section 4.4) andwill lead to the detection of lower crystallization temperaturesupon cooling than the use of slow cooling rates.

In conventional DSC instruments calibration of the temper-ature scale and the determined heat of transition is carried out ona regular basis usually using reference substances of high purity

[50]. The most common calibration substance is indium (m.p.156.6 °C) which is suitable for temperature as well as for heat(peak area) calibration. Other metallic substances like tin (m.p.231.9 °C), lead (m.p. 327.5 °C) or zinc (419.5 °C) can be usedfor the generation of additional calibration points. Unfortunate-ly, the comparatively high melting point of most metals is out ofthe temperature calibration range of highest interest for theinvestigation of aqueous lipid dispersions. As the only relevantexception is gallium (m.p. 29.8 °C) different organic materialsor water (m.p. 0.0 °C) may be included in the calibrationprocedure to cover the whole temperature range relevant for thestudy of lipid nanoparticles. For accurate measurements,calibration should be performed at the same scan rate asemployed for the measurements (unless a significant influenceof the scan rate on the results obtained is ruled out bycomparative investigations at different scan rates).

4. Applications of DSC and X-ray diffraction for thecharacterization of LNP dispersions

4.1. Some considerations on sample preparation

As outlined inmore detail below, LNPs often have very peculiarproperties which are due to their dispersed state and the smallparticle size. In order to retain these properties during investigationcare has to be taken not to induce alterations by sample preparationprocedures. If at all possible the dispersions should thus beinvestigated in their native state.Modern sensitiveDSC instrumentsare capable of detecting the crystalline lipid phase transitions at thetypical concentrations in LNP dispersions of about 1–10% or evenbelow. Sensitivity may become an issue e.g. in the investigation ofminor compounds of the particle composition or for phasetransformations which involve small amounts of energy (forexample, liquid crystalline phase transitions). Moreover, investiga-tions of highly diluted samples can be problematic in particular alsofor X-ray investigations. If some kind of sample modification isinevitable prior to investigation, e.g., in the case that very diluteddispersions have to be concentrated to overcome problems withinstrument sensitivity, the particle properties must not be changedby the sample preparation procedure, especially in terms of physicalstate and particle size. Against this background, drying of sampleswhich is sometimes performed for analytical purposes isparticularly critical. Drying may lead to pronounced alterations ofsample properties as result of temperature effects (particularly uponfreeze-drying) but also simply due to the loss of the dispersionmedium. The effects may, for instance, concern transitiontemperatures, melting enthalpy and polymorphism as, e.g. reportedin [48,57–59] and illustrated in Fig. 14. In cases where drying isinevitable the potential influences of the sample preparationprocedure on the results obtained need to be carefully consideredupon data interpretation.

4.2. Observation of general thermal phenomena in LNPdispersions

Probably the most obvious, easiest and most frequently usedapplication ofDSC in the study of lipid nanoparticle dispersions is

Fig. 14. Effect of air drying on the thermal behavior of triglyceride dispersions. Left: Trilaurin dispersion (10% trilaurin, 6% tyloxapol in an aqueous phase containing2.25% glycerol and 0.01% thiomersal) prepared by melt homogenization. The original sample is an emulsion of supercooled liquid droplets which do not lead to athermal event upon heating. After air drying an equivalent amount of dispersion at room temperature the resulting thin film displays a pronounced melting event. Right:Tripalmitin dispersion (10% tripalmitin, 10% tyloxapol in an aqueous phase containing 2.25% glycerol and 0.01% thiomersal; scan rate 2.5 °C/min). The originaldispersion displays a structured melting event due to its small particle size (PCS z-average 60 nm, polydispersity index 0.20). The structuring is completely lost after airdrying the sample.

Fig. 15. Small and wide angle X-ray diffraction patterns of cholesteryl myristatein bulk and in colloidal dispersion (5% cholesteryl myristate stabilized with 5%polyvinyl alcohol). Bulk: crystalline powder (a) at 20 °C and smectic mesophase(b) at 60 °C. Dispersions stored at 4 °C (c) and at 23 °C (d) and measured at20 °C. The graphs of the bulk material and the dispersions are not on the same(linear) intensity scale. The smectic phase is characterized by a pronouncedsmall angle reflection and the absence of wide angle reflections whereas thecrystalline material displays reflections both in the small and in the wide anglerange (from [64] with permission).


the investigation of the physical state of the dispersed particles viathe detection of characteristic melting or transformationendotherms upon heating (or the absence of these endotherms,respectively).Moreover, the thermal behavior of the nanoparticlesupon cooling is also conveniently studied with this method andthe results may provide valuable information for the layout ofpotentially required thermal treatment after processing stepsinvolving heat treatment [43,46,47,60]. The necessity to payspecial attention to the physical state of LNPs arises from the factthat the properties of the lipid matrix materials may changesignificantly when they are brought into the colloidally dispersedstate. For example, even when crystalline lipids are used for thepreparation of LNP dispersions, the lipid particles are notnecessarily present in the solid state after processing, in particularwhen the preparation process is carried out in the heat such asduring melt homogenization [46,47,58,60–64]. This is due topronounced supercooling in the dispersed state, a generalphenomenon not only observed in lipids [61,65]. Considerablesupercooling of the melt prior to crystallization is already oftenobserved in bulk lipids, such as in triglycerides or cholesterolesters [47,64]. In nanodispersed lipids, due to the absence ofparticulate impurities (which promote crystallization of the bulkmaterial) in the vast majority of the droplets crystallization isusuallymore retarded and occurs at lower temperatures than in thebulk lipids [9,47,60,61,64,65]. For instance, colloidally dispersedsaturated monoacid triglycerides may crystallize at temperaturesabout 20 °C lower than the bulk material leading to a differencebetween the melting maximum and the onset of crystallization ofup to almost ∼50 °C [47,49,60,66]. In other types of lipidmaterials, for example complex glyceride mixtures or paraffins,the supercooling effect is, however, less pronounced [67–69].The confirmation of the desired physical state of thematrix lipid isthus of crucial importance for the development of nanoparticlesbased on solid lipids. In most cases, the desired state of theparticles will be solid or at least partially solid (in some oil-containing solid LNPs). For special applications, however, also

the state of a supercooled liquid melt may be aimed at, forexample to achieve particular carrier properties [61] or to increasethe bioavailability of highly lipophilic drugs [70]. Recently, alsothe formulation of nanoparticles based on supercooled smecticliquid crystals consisting of cholesterol esters has been proposed

Fig. 16. DSC heating curves of trimyristin (D114) nanoparticles illustrating theeffect of particle sizes (the approximate PCS z-average particle sizes [nm] aredisplayed at the curves) on the shape of the melting curve. For comparison, theheating curves of more complex triglyceride materials (the hard fats Softisan 154(S154; containing approximately 49% C16 and 46% C18 chains) and WitepsolH42 (H42; containing approximately 44% C12, 15% C14, 23% C16 and 16%C18 chains)) are also given. For particles of comparable size the structuring ofthe melting event decreases with increasing complexity of the matrix material.

Fig. 17. Transmission electron micrograph of a freeze fractured specimen of atrimyristin dispersion (stabilized with 2.4% soybean lecithin and 0.6% sodiumglycocholate) (top) and schematic representation of the arrangement of thesingle molecules in a triglyceride crystal in the β-form (bottom; modified after[72]). The platelet-like shape and molecular layers of the triglyceridenanoparticles are clearly visible in the micrograph. The small sphericalstructures represent vesicles formed by excess phospholipid.


as a new type of colloidal lipid carriers [64]. All these dispersionsare characterized by the presence (in the case of crystalline orliquid crystalline particles) or absence (supercooled melts) ofcharacteristic thermal transitions in the DSC heating curveallowing to draw conclusions on their physical state. Theseconclusions may be confirmed by additional X-ray investigationswhich are particularly helpful in the case of liquid crystallinenanoparticles (Fig. 15) the thermal transitions of which can bevery small. Simultaneously, information on the position of themelting and crystallization transitions can be collected. Suchinformation is often required for a rational layout of thepreparation process, in particular, to determine adequate coolingconditions to crystallize the particles from their melted state (or,on the other hand, to find out lower critical temperatures for thedevelopment of stable supercooled systems).

4.3. The effect of the small particle size on the thermalbehavior

When comparing DSC thermograms of bulk lipids andcorresponding solid LNP dispersions differences in shape andposition of the signals are usually observed: The endotherm ofthe nanoparticles is broadened and shifted to lower temperatures(Fig. 12). In addition to possible effects of the adsorbedemulsifier molecules, an effect of the particle size on the meltingtemperature is to be expected for all nanodispersed materialsaccording to the Gibbs–Thomson equation [71]:

lnTT0

¼ −2gslVS

rDHfusð11Þ

T Melting temperature of a particle with radius rT0 Melting temperature of the bulk material at the same

external pressure

γsl Interfacial tension at the solid–liquid interfaceVS Specific volume of the solidΔHfus Specific heat of fusion

This effect is reflected in a decrease in melting temperaturefor a particle of given size compared to the bulk material andbecomes particularly pronounced in the lower nanometer sizerange. Because LNP dispersions are usually more or lesspolydisperse the melting transition is not only shifted to lowertemperatures but is also broadened since the fractions ofdifferent particle sizes melt at different temperatures. Fordispersions containing particles with fractions of specific, butdistinctly different dimensions even the occurrence of a series ofsharply separated melting events can be observed. Suchbehavior has been described for small size dispersions ofsaturated monoacid triglycerides in the stable β-modification[29,30,33,49] (Fig. 16, cf. Section 2.3.1). These particlesrepresent thin platelets which can only exist in thicknessescorresponding to the multiples of a single molecular triglyceridelayer (Fig. 17). As the thickness of a triglyceride monolayer(e.g., 4.06 nm for the β-form of tripalmitin) is relatively large


compared to the overall thickness of the triglyceride nanoplate-lets (which may consist of only several molecular layers) thesteps in particle thickness are quite pronounced. The thicknessdistribution of the particles then leads to the occurrence of astructured melting event in which each peak represents themelting of a fraction with specific particle thickness [29]. Thephenomenon becomes less pronounced with increasing com-plexity of the dispersed material, probably since complex lipidmixtures are unable to form particles with defined gradations inthickness thus producing only a diffuse, broad melting eventwith comparatively low transition temperature [30] (Fig. 16).

Other effects observed with decreasing particle size incolloidal solid triglyceride dispersions were a decrease inmelting enthalpy (as also observed by [66]), an increase in therate of polymorphic transitions as well as the occurrence ofunusual reflections in small and wide angle X-ray diffracto-grams. Although a small decrease of the crystallizationtemperature with decreasing particle size has also been reportedfor colloidal triglyceride nanoparticles this effect was much lesspronounced than the influence on the melting temperature(smaller than 2 °C for nanoparticles with PCS mean particlesizes between about 65 nm and 500 nm) [30].

The particle size related broadening and structuring of themelting transition in LNP dispersions significantly complicatesthe interpretation of the DSC curves. As found in investigationson the behavior of triglyceride nanoparticles, in certain cases thedepression of the melting temperature due to the particle sizeeffect may be so large, that the melting transition could beconfused with that of a lower melting polymorphic form [30,60].The interpretation of melting curves of lipid materials with morecomplex composition (which can already be very demanding forthe bulk state) will be particularly complicated if pronouncedinfluences of the particle size occur. In addition, the broadeningof the DSCmelting endotherms in nanoparticle dispersions oftenimpairs the correct integration of the signals (as does theirgenerally small size and sometimes also wide melting range dueto other effects like complex composition and/or polymorphicbehavior of the matrix materials). Larger errors compared toinvestigations on bulk materials have, therefore, to be taken intoconsideration upon quantitative evaluation.

Fig. 18. Schematic representation of the molecular arrangement in the different tri[10,80,81]).

As a result of the particle size effect, it is often difficult todefine a characteristic melting point of the nanodispersematerials. The melting endotherm reflects melting of a hugearray of particles each of which behaves individually (e.g., a10 mg DSC sample of a 10% glyceride dispersion with a particlesize of 150 nm contains in the order of 1011 nanoparticles). Inprinciple, one would need to define the melting point of eachsingle particle which is, of course, impossible in practice. Anextrapolated onset value can often not be determined from themelting curves in a useful way since the shape of the curvesdeviates quite significantly from that of a ‘neat’ signal of a purebulk substance unless the particles are rather coarse. As analternative, the peak maximum temperature is often used. Thisparameter has the advantage of being also accessible forcomplex melting events with shoulders or side maxima,resulting, e.g., from polymorphic transitions or melting infractions. A potential influence of the experimental conditionson the peak maximum temperature needs, however, to be takeninto consideration (cf. Section 3). For very complex curves, e.g.for those with several maxima, the assignment of a singlecharacteristic temperature does not appear to be useful and otherparameters (positions of different peaks, span of the transition ordisplay of the whole curve) may be more appropriate.

4.4. Polymorphism and aging phenomena

An important point to be considered in the characterization ofLNP dispersions is the potential occurrence of differentpolymorphic forms. Polymorphism is commonly found in lipidmaterials and thus also in the components used in lipidnanoparticle dispersions, for instance, glycerides or fatty acids.The different molecular packing in the polymorphic modifica-tions is reflected in differences of physical properties like themelting points and enthalpies. For simple lipids, like the saturatedmonoacid triglycerides or fatty acids, the properties of thedifferent forms in the bulk are well-known [73–79] and can alsobe found in the nanodispersed materials. For other lipid materialslike, e.g., complex glyceride mixtures (e.g., Witepsol, Softisan,Compritol) or waxes (e.g., cetylpalmitate), which are alsofrequently employed as matrix materials in LNP dispersions,

glyceride polymorphs (top) and of the resulting XRD patterns (modified after

Fig. 19. Schematic representation of the B and C form of fatty acids (modifiedafter [76]).

Fig. 20. Influence of heating rate on the shape of the melting curve of a freshlycrystallized tristearin dispersion (10% tristearin, 2.4% DPPC, 0.6% sodiumglycocholate) and the ratio of the polymorphic forms detected. The metastableform originally present in the sample can only be reliably detected at highheating rate. At low heating rates, transformation into the stable β-form occursduring the heating run. The curves have been normalized with respect to theheating rate; the structuring of the β-transition is due to particle size effects.


published specific reference data is rare or even unavailable. Forsuch materials (but sometimes also for simple lipids due to theparticle size related distortion of the melting transition) the correctassignment of the polymorphic forms in DSC curves can be amajor challenge. A sound interpretation of more complex meltingcurves thus requires information from complementary investiga-tions, in particular by X-ray diffraction which allows moreunambiguous assignment of the polymorphic forms and may alsohelp to detect processes such as phase separations.

Glycerides occur in three major polymorphic forms, α, β′and β (in order of increasing thermodynamic stability) whichare characterized by different subcell packings of the lipidchains, different angles of tilt of the lipid chains with respect tothe molecular glyceride layers and different densities. Thedifferent subcell packings lead to characteristic wide angle X-ray reflections enabling their identification (Fig. 18). Forsaturated monoacid triglycerides the α- and β-forms arefrequently observed in the bulk state as well as in colloidaldispersions [31,46,81,82] whereas the β′-form usually occursonly under special conditions [84]. The β′-modification is,however, frequently observed in more complex triglyceridesand triglyceride mixtures and in compositions containing largerfractions of partial glycerides (e.g., in many hard fats or glycerylbehenate) where it often represents the storage stable form[46,48,85]. The presence of the β′-form also in colloidalparticles has been confirmed by X-ray diffraction on glycerylbehenate nanoparticles [44,48,68]. Similar wide angle diffrac-tion patterns were also found for nanoparticles containing solidparaffin and cetyl palmitate [45]. Moreover, for complextriglyceride mixtures like hard fats an intermediate β-form(βi) has been described [86,87] which has also been observed incorresponding LNP dispersions [82,46].

The polymorphism of the major glyceride modifications is ofthe monotropic type, i.e. the form with the highest melting pointis thermodynamically stable over the whole temperature range.Metastable polymorphs may thus transform into a more stableform already below their respective melting point, e.g., simplyon storage. For instance, in saturated monoacid triglycerides themetastable α-modification is usually formed upon crystalliza-tion from the melt. Transformation into the stable β-form cannotonly be induced by heating the system to the melting point ofthe α-modification but also occurs below this temperature uponaging of the sample. The kinetics of polymorphic transitionfrom the α- into the β-form below the α-melting temperature

depend on the type of triglyceride (shorter-chain triglyceridesundergo the transition more easily) and the storage temperature(storage temperatures more close to the α-melting point willlead to more rapid transitions). These principles, which wereoriginally established for bulk triglycerides [88] have also beenobserved in triglyceride nanoparticles [47,83].

LNPs prepared on the basis of fatty acids have lessintensively been studied with regard to their solid state structurethan glyceride based nanoparticles. Even saturated fatty acidslike stearic and palmitic acid which are usually employed for thepreparation of the corresponding systems can exist in fourdifferent polymorphic forms termed A, B, C and E [78,79,89].All forms can be obtained upon crystallization from solution butonly the C form grows upon crystallization from the melt. Thisform has also been found with DSC and X-ray diffraction in(dried) stearic acid nanoparticle preparations [90–93]. Althoughthe B form is the stable modification of stearic acid below 32 °C[78,79] this form has only rarely been found in nanoparticulatematerial based on this substance, for example as a result ofinteraction with alcohol cosurfactants or drug incorporation[90–93]. The structure of the B and the C form is quite similar.Both forms have an orthorhombic subcell (similar to the β′-form of glycerides) and mainly differ in their angle of tilt(Fig. 19). They can be distinguished by their different meltingpoints (with that of the C form being higher). In X-raydiffraction, the two forms can be identified by differences in thelong spacings (corresponding to the distances between thesingle molecular layers) ([90,91,94]) whereas the wide anglereflection patterns for both forms are very similar due to thesame type of subcell packing.

With DSC, the polymorphic status of a sample can bechecked and the influence of different parameters such asstorage time and temperature or interactions between thecomponents of the dispersion can be studied this way. Wheninvestigating the presence of metastable polymorphs it needs tobe taken into consideration that the sample properties may be


altered by the heating program of the thermal analyzer in thecase of monotropic phase behavior of the matrix lipids.Metastable polymorphs may thus transform into more stableones during analysis. Such transformations are sometimesreflected by distinct recrystallization events in the DSC curvebut can also proceed very gradually and may then escapeobservation (Fig. 20). Obviously this would lead to anunderestimation of the presence of metastable form. As gradualtransformations are particularly observed at slow heating rates,high heating rates should be used to limit this problem in theanalysis of polymorphic samples. Unfortunately, using higherheating rates will decrease resolution of multiple signals so thata compromise may be necessary. Methods operating at constanttemperature like X-ray diffraction are advantageous in thisregard. Although in principle well suited for this purpose, X-raydiffraction has so far not evolved into a routine method to studygradual polymorphic alterations of lipid nanoparticles in aquantitative way, probably because quantitative data is moredifficult to obtain by this method, particularly for aqueousdispersions.

The correct, quantitative determination of the ratios of thedifferent polymorphic forms by DSC is often difficult inpractice due, for instance, to the superimposition of thermalevents, particularly for complex matrix compositions. In suchcases, following simply the evolution of the overall transitionenthalpy in dependence on the parameters of interest, e.g., independence on storage time, may be a pragmatic solution[83,95]. Often, this enthalpy is normalized to the meltingenthalpy of the thermally untreated raw material (taking thereduced concentration in the dispersion into account) and thenreferred to as ‘recrystallization index’ or ‘(degree of) crystal-linity’, usually given in % [57,60,66,85,96–99]. Although thisprocedure provides data which can be compared more easily, forexample when different matrix materials are under investiga-tion, some points need carefully to be considered upon its use.In a strict sense, only the transformation of amorphous intocrystalline material (as, e.g., occurring upon cooling after

Fig. 21. Wide angle X-ray diffractograms of tristearin systems. Top: Thermallyunstressed raw material; middle: raw material recrystallized from the melt andstored for 6 days at room temperature; bottom: dispersion (10% tristearinstabilized with 2.4% soybean phospholipid/0.6% sodium glycocholate) storedfor 6 days at room temperature. The curves of the bulk materials and thedispersion are not on the same intensity scale (s=1/d=sin θ /λ; 2θ: scatteringangle; λ: wavelength (0.15 nm)) (from [95]).

preparation of melt-homogenized samples or during storage ofdispersions of supercooled droplets under inadequate condi-tions) should lead to an increase in ‘crystallinity’. In contrast, anincrease in melting enthalpy due to polymorphic transitionsshould not be considered as reflecting an increase incrystallinity as the less stable polymorphic forms are alsocrystalline even though their melting enthalpy is lower. Even ifthe fully crystallized raw material in the stable polymorphicform is regarded as the reference for normalization, relevantreference values are sometimes difficult to obtain. Polymorphictransitions generally occur more rapidly in lipid nanoparticlesthan in the bulk state [49,66,82,95] as illustrated in Fig. 21.Some materials (like certain hard fats) even transform into amore stable polymorphic form than is observed in the bulkmaterial when they are dispersed into colloidal particles asconfirmed by X-ray diffraction [46,66]. Also glyceryl behenatenanoparticles have been observed in a more stable form thantheir bulk material under certain conditions [48]. In such cases,using the melting enthalpy of the raw material as referencevalue may obviously lead to a ‘degree of crystallinity’ higherthan 100% in the nanoparticles. In some cases, the crystalliza-tion index does not refer solely to the crystallizing lipid but tothe whole matrix lipid used for particle preparation, even if itcontains lipid material that is not expected to crystallize in theparticles, e.g. lecithin [66,85]. This procedure leads to maximal‘recrystallization indices’ below 100%. Low values may alsoresult from the fact that the melting enthalpy in small LNPs islower than that of the bulk material. This has been attributed tothe contribution of interfacial effects [30]; a potential influenceof the emulsifier layer on the enthalpy values has not yet beeninvestigated. Moreover, for an accurate determination of therecrystallization index with reference to the heat of transition ofthe bulk material the exact concentration of matrix lipid in thedispersion has to be known in order to provide a correct basisfor normalization. Although the concentration of matrix lipid inthe processed dispersions may deviate from that of the primarycomposition used for preparation due to, e.g., evaporation ofwater during processing in the heat or dilution with processwater in some homogenizers, determination of the lipid contentin the final formulations is usually not performed. In somespecial cases, however, a reasonable estimate of the fraction ofcrystalline material may be possible even without knowledge ofthe sample concentration. For example, an approximation of thefraction of recrystallized matrix material in dispersions ofsupercooled melts or liquid crystals can be obtained by relatingthe melting enthalpy of this recrystallized fraction not to that ofa bulk sample but to that of the same dispersion after enforcedrecrystallization of the dispersed materials, for example bycooling to very low temperatures [64,100]. This procedurerequires that enforced crystallization leads to the directformation of nanoparticles in the fully crystallized state withoutformation of metastable polymorphs. Even when the materialcrystallizes into stable polymorphic forms under these condi-tions, freshly crystallized materials often have a lower meltingenthalpy than after a certain time of storage. In these cases, useof this procedure leads to a slight overestimation of therecrystallized fraction [64]. One possibility to overcome this


problem could be the use of an artificially crystallized referencesample brought into the final state by aging from a fraction ofthe dispersion of interest.

A gradual increase in melting enthalpy over, e.g., severaldays or weeks of storage is not uncommon in dispersions ofsolid lipid nanoparticles as a result of polymorphic transitionsand the formation of a higher degree of crystalline order[60,83,95–97]. Such processes are of high practical relevancesince there are indications that the drug incorporation capacityof glyceride nanoparticles decreases with increasing packingorder of the matrix lipids [46,48]. Moreover, alterations in solidstate properties lead to changes also in other sample properties,e.g. the particle shape [31,95] which, in turn, might alter therheological properties [32,101]. Also the occurrence of colloidalinstabilities in dispersions of glycerol behenate nanoparticlescould be related to an increase in melting enthalpy and alteredshape of the melting curve [96]. Jenning et al. reported onpolymorphic transitions in glyceryl behenate nanoparticles dueto the presence of electrolytes and under the conditions of aFranz cell drug release experiment as observed by DSC and X-ray diffraction [48].

4.5. Influences of dispersion composition

4.5.1. Dependence of the particle properties on the type ofemulsifier

Emulsifiers used for dispersion preparation can influence theproperties of the dispersed matrix lipids with regard to theircrystallization behavior in different ways as established by DSCand XRD studies. In a screening of triglyceride nanodispersionsstabilized with a large number of different surfactants it wasfound that there is a critical lower crystallization temperaturerange for a given triglyceride material in which the nanopar-ticles in most of the investigated dispersions crystallized [102].This result is in good agreement with the assumption thatcrystallization in the dispersions occurs by homogeneousnucleation. Stabilization with certain surfactants leads, howev-er, to an increase in crystallization temperature by severaldegrees. A closer investigation of this phenomenon [102]reveals that the surfactants leading to increased crystallization

Fig. 22. Time dependent evolution of thewide angleX-ray diffraction patterns of trimyriconsisted of 10% trimyristin stabilized with 2.4% phospholipid (either soybean lecithin (The time points given in the figure refer to the first appearance of a small angle reflesmall reflection in the diffractogram of the DPPC-containing dispersion at 19.5 °C is d

temperatures contain comparatively long, saturated alkyl chainswhereas surfactants leading to crystallization in the lowercritical temperature range possess short, branched or unsaturat-ed lipophilic residues. The same principles also apply to the useof phospholipids as components of stabilizer mixtures fortriglyceride nanoparticles [83]. The influence of some surfac-tants on the crystallization temperature is attributed to theinduction of heterogeneous nucleation due to interaction of thelipid melt with the surfactant layer. The effect may be due to atemplating effect of the orientated surfactant chains or even tothe formation of a crystallized surfactant monolayer on thesurface of the lipid particles as indicated by a ‘precrystallization’event prior to the main triglyceride crystallization transitiondetected by DSC and X-ray diffraction investigations in most ofthe corresponding dispersions [83,102]. Influences of the typeof surface active components on the crystallization temperatureshave also been reported from investigations in food science[8,103–106]. In addition to the influence of the emulsifier onthe crystallization temperature effects on the kinetics of poly-morphic transitions were reported in studies on triglyceridenanoparticles [31,60,83,97,102]. The effects are related to thetype of hydrophilic headgroup [102] but also to the structure ofthe lipophilic chain with long, saturated acyl chains delaying thetransformation into the stable polymorph in comparison withshorter or unsaturated chains [83]. Besides DSC, time resolvedX-ray diffraction studies are very helpful to study such kineticphenomena, in particular for rapidly transforming systems(Fig. 22). Stabilization of triglyceride nanoparticles solelywith bile salts like sodium cholate or sodium glycocholatealso delays the transition into the stable form significantly[31,60,97,102]. Moreover, the sole stabilization with sodiumglycocholate has been found to induce a pronounced depen-dence of the internal particle structure of the metastable α-form,in particular the formation of ordered molecular triglyceridelayers, on the crystallization conditions as reflected in DSC andXRD results [31].

For cholesteryl myristate nanoparticles in the supercooledsmectic state effects of the emulsifiers on the stability againstrecrystallization and on the course of the crystallization eventupon cooling have been observed by DSC. Fatty acid chain

stin nanoparticles during crystallization from themelt upon cooling. The dispersionsS100) or dipalmitoylphosphatidylcholine (DPPC)) and 0.6% sodium glycocholate.ction which was considered to mark the onset of triglyceride crystallization. Theue to the solidified chains of the phospholipid (from [83] with permission).


containing surfactants (e.g. phospholipids, fatty acids, sucroseester) usually lead to a multiple crystallization event and todispersions less stable against recrystallization of the lipid matrixthan typical polymeric surfactants (e.g. poloxamer, polyvinylalcohol) [107]. The occurrence of different polymorphic formscould be ruled out as the cause of the complex crystallizationevent in phospholipid/sodium glycocholate stabilized disper-sions. Instead, the effect seems to be related to the presence ofdifferently shaped particles in these dispersions [100].

4.5.2. Use of lipid mixtures as matrix materialsMainly mixtures of two specific components for lipid

nanoparticle preparation will be considered at this point becausethe influence of single components present in themulticomponentmixtures often employed as prefabricated matrix material forLNPs (e.g. hard fats) has hitherto only scarcely been studied. Theadmixture of a second glyceride to a given matrix material hasbeen observed to lead to alterations in crystallization and meltingbehavior as well as polymorphic transitions. The crystallizationtemperature of the nanoparticles can either be increased ordecreased by the admixture of a second lipid as observed in DSCmeasurements. For example, the addition of tristearin (C18-chains) to trilaurin or trimyristin (C12- or C14-chains, respec-tively) decreases the supercooling tendency of the shorter-chaintriglycerides since the long-chain triglyceride induces nucleationat higher temperature [47]. Decreased supercooling has also beenobserved in mixed-chain triglyceride hard fats compared to amonoacid triglyceride with similar average chain length [46]. Onthe other hand, the addition of liquid, medium chain triglyceridesto glycerol behenate reduces the crystallization temperature of thenanoparticles [68,108]. A reduced crystallization tendency uponstorage was also observed for hard fat nanoparticles afteradmixture of medium chain triglyceride oil [99]. The crystalli-zation tendency of supercooled smectic cholesterol esternanoparticles could be modified by admixture of a second esterwith lower crystallization tendency [107].

For glyceride nanoparticles prepared from mixtures of solidand liquid lipids also a depression of the melting point of thesolid lipid has been reported with increasing concentration ofliquid lipid [44,68,108,109] which clearly indicates theassociation of the two lipids within the single nanoparticlesalso after crystallization. The kind of interaction of the twomatrix components cannot, however, be straightforwardlyderived from the DSC data for this type of mixture. A meltingpoint depression can, in principle, be caused by the formation ofa solid solution or by eutectic behavior between liquid and solidlipid. For the glyceryl behenate/medium chain triglyceridemixtures under investigation in the mentioned studies a slightsolubility of the liquid component within the solid matrix may beexpected. The melting point depression has, however, mainly tobe attributed to eutectic behavior since the results of othermethods point to intraparticulate phase separation of the liquidfrom the solid lipid already at quite low oil concentrations whichleads to oil localization at the surface of the solid lipid[23,24,44]. The presence of enclosed liquid oil compartmentswithin a matrix of solid lipid has often been proposed fornanoparticles consisting of solid/liquid lipid mixtures [[68,108–

111]. There are, however, so far no convincing experimental datasupporting this concept. A melting point depression of the solidcompound due to admixture of a liquid component has also beenreported for freeze-dried mixed tripalmitin/medium chaintriglyceride and stearic/oleic acid nanoparticles prepared inanalogy to nanoparticles from solid/liquid glyceride mixtures[110,112]. For freeze-dried formulations conclusions on thestate of the mixture within the original nanoparticles are,however, even more difficult to draw than for the dispersionsince the structure of the mixture may change during the dryingprocedure.

Also for binary mixtures of saturated monoacid triglyceridesin the stable β-form, eutectic behavior would be expected inmost cases due to the limited intersolubility of triglyceridesunder these conditions [81,113,114]. The extraction of phasediagrams of binary mixtures from DSC data is, in principle, astandard procedure (not only for lipid materials) which shouldallow assignment of eutectic/monotectic behavior, compoundformation or the formation of solid solutions [115–117]. In aninvestigation on the melting behavior of a series of mixedtrilaurin/tristearin and trimyristin/tristearin nanoparticles theinterpretation of the DSC heating curves was, however, difficult.Complications arose from the very broad melting events, oftenwith the occurrence of several additional peaks [47]. Bothphenomena can at least partially be attributed to the particle sizeeffect. This effect is hard to distinguish from an extendedmeltingrange expected for the higher melting compound in the twophase region of the eutectic phase diagram andmakes the correctassignment of the major thermal events (melting of eutecticmixture and liquidus point) difficult. Although some conclu-sions were possible on the melting behavior of thesenanoparticles (particularly the confirmation that the singleparticles still contained the triglyceride mixture) the phasebehavior could not unambiguously be clarified in detail evenwith the help of additional X-ray data. X-ray diffraction did,however, allow to clearly distinguish a trilaurin and a tristearin-rich component in the respective nanoparticles and indicated ahigher solubility of trilaurin in tristearin than vice versa.

Concerning the influence on polymorphic transitions, anenhancement of the transition into a more stable form has beenobserved by DSC and X-ray diffraction for glyceryl behenatenanoparticles after addition of liquid medium chain triglycerideoil [68]. This is in agreement with investigations on bulkmaterial where increased transition rates have been reported forglycerides after addition of liquid oils [118]. Effects on the timecourse of polymorphic transitions have also been observed innanoparticulate mixtures of the two solid triglycerides tristearinand trilaurin [47].

Recently, the admixture of lecithin has been proposed to alterthe matrix properties of solid hard fat nanoparticles [66,85,119].The melting and crystallization properties of the resultingnanoparticles were, however, still mainly governed by the hardfat component [66,85]. A more detailed investigation of theultrastructure of these dispersions, amongst others with the aidof SAXS, pointed to the formation of a triglyceride-richnanoparticle core surrounded by a shell rich in phospholipidsrather than to the presence of lecithin in the particle core [120].


4.5.3. Investigations on the effect of drug loadingDSC and XRD are often used to examine the effect of drug

loading on LNPs and to evaluate the state of the drug (crystalline,amorphous, molecularly dispersed) in the dispersions. Whenapplying these techniques for the investigation of the state ofincorporated drugs it should be kept in mind, however, thatneither method is designed to find minor amounts of foreigncrystalline substances in a carrier matrix. In X-ray diffraction, thedetection limit depends much on the relative intensity andposition of the signals of the component of interest compared tothat of the carrier matrix (as well as stabilizers and otherdispersion components). Even the occurrence of comparativelylarge signals arising from crystalline drugwill be difficult to detectin a straightforward way when they are superimposed by thesignals of the carrier matrix. At low drug concentrations, thesignals may be completely lost in the baseline noise. Peak fittingprocedures, e.g. by Rietveld analysis [121], may help to identifydifferent crystalline components even in case of superimposedreflections but have hitherto not been used in the field of LNPresearch. As a first approach to evaluate whether the presence ofcrystalline drug (or other components of the mixture) would bepossible to detect in the respective dispersions mixtures of thecomponents in the same quantitative composition as present in thedispersion may be measured for comparison [43,122]. Thisapproach still does not consider the fact that the signals fromcolloidal materials are broader and thus more difficult to analyzethan those of bulk substances. As a further complication, thesignals from dispersed material are generally less intense thanthose from bulk substances due to the low concentration of theparticles in the dispersion medium (Figs. 15 and 21). Conse-quently, the samples are often dried prior to analysis which may,however, produce artifacts and cause further complications, e.g.,due to the presence of cryoprotectants [122].

In DSC, the situation is even more complicated since thephysical properties of the sample change during the temperatureprogram. Therefore, the positions of the signals cannotnecessarily be expected at the same positions as those of thepure raw materials. This will only occur when there is nomiscibility of the corresponding substances, neither in the solidnor in the melted state. Usually, however, drugs incorporated intoLNPs are soluble in the lipid melt so that an influence on thetransition temperatures has to be expected, e.g., due to eutecticbehavior. Drugs which were incorporated in the matrix lipid viadissolution in the melted lipid and are still associated with theparticles after preparationmay thus not give rise to a signal at theirbulk melting point since, at this temperature, they may haveredissolved in the lipidmelt (very high heating rates may suppressthis process to some extent). If at all, the melting transition of thedrug will then be detected at lower temperatures. The detection ofvery small amounts of crystalline material can be difficult due tolow concentration of drug as well as large width and thesuperimposition of transitions. The investigation of mixturesof the raw materials in the same ratio as used in the disper-sions (as reported in [43,91,122–124]) may help to check forcorresponding effects. Although the absence of drug meltingtransitions as, e.g. observed in [43,123,124], may point to apreferably amorphous or molecularly dispersed state of the

incorporated drug the absence of clear drug melting events inDSC curves cannot be taken as final proof of a non-crystallinestate of the drug in the sample. While the detection of acorresponding endothermic transition in an aqueous dispersionwould be an unambiguous indication for the presence ofcrystalline drug (to the knowledge of the authors this has notyet been reported), the situation is less clear for dried samples (inwhich drug crystals might be formed during drying). Anyway, theoccurrence of such melting signals (which have only scarcelybeen reported [93]) is usually taken as representative also for thesituation in dispersion. In many cases it will be impossible toinvestigate the native aqueous dispersions for the presence ofcrystalline drugs asmany of the substancesmelt above 100 °C andthese temperatures are difficult to achieve in DSC measurementsof aqueous samples. At the moment, drying of the dispersionsseems to be the only alternative but the effects of this type ofsample preparation remain to be investigated before itsapplicability can finally be evaluated.

An important question after drug incorporation into LNPdispersions is whether the drugs incorporated are indeedassociated with the nanoparticles. Such an association may bereflected rather indirectly by an influence on the phasetransitions of the matrix lipid. Although often regarded asconfirmation of drug incorporation, the depression of themelting point of the matrix lipid is not necessarily an indicationof drug incorporation into the lipid matrix or even into thecrystalline lattice. For example, the presence of ubidecarenone,which, at higher drug loads, forms a separate liquid phase ofsupercooled drug material within single nanoparticles, causes aconcentration dependent melting point depression of the solidlipid in triglyceride nanoparticles due to eutectic behaviorbetween the two substances [43]. The melting point depressionof cetyl palmitate in nanoparticles loaded with a high fraction ofthe liquid active tocopheryl acetate [125] is probably also due tothis phenomenon. Although a depression of the melting point oflipid nanoparticles at lower (model) drug loads has sometimesbeen reported [43,57,93,122,126] small effects are difficult todetect in the dispersions as they may be superimposed, e.g. bythe particle size effect. Moreover, the drug load in LNPs is oftenmuch below 10% so that distinct effects on the melting pointcannot be expected. In addition to their influence on the meltingtemperature high loads of liquid substances may also result in adecrease of the melting enthalpy due to dissolution of fractionsof the solid lipid in the liquid phase [43]. Effects on the meltingenthalpy have, however, also been reported for lower drug loads[57,91,93]. Moreover, the presence of liquid or non-crystallizingactive substances like ubidecarenone or tocopheryl acetate hasbeen found to decrease the crystallization temperature ofglyceride nanoparticles in a similar way as the loading with oil[43,127]. For smectic cholesteryl myristate nanoparticles adecrease in the crystallization temperature was observed byDSCafter loading the dispersions with 10% (related to the matrixlipid) of ibuprofen, etomidate and miconazole [107]. For thistype of particles, also a depressing effect of the drugs on theliquid crystalline phase transitions upon heating was observedwhich further confirms the association of drug with thenanoparticles.


With regard to polymorphic transitions, XRD revealed anenhancing effect of ubidecarenone on the transition oftripalmitin nanoparticles into the stable β-modification. Someeffects of the presence of other drugs on the polymorphictransitions on triglyceride nanoparticles [127] as well as ofmodel substances on the polymorphic behavior of dispersedstearic acid [91,93] have also been reported.

4.6. Solid LNPs in semisolid formulations

Although most DSC and X-ray characterization work on LNPshas so far been done on the primary dispersion formulations (ortheir dried analogues) some investigations on further processednanoparticles have been reported. These studies mainly concernsolid LNPs in semisolid formulations. Using DSC, the preparation-dependent distribution of the matrix lipid of paraffin nanoparticlesin the different phases of a cream formulationwas investigated [67].By the same method, the presence and solid state of cetyl palmitatenanoparticles in gel and cream formulations was confirmed[125,128]. Also by DSC, an increased crystallization tendency ofhard fat nanoparticles after incorporation into hydrogel formula-tions compared to the native dispersions was reported [99]. Jenninget al. [48] used X-ray diffraction to confirm the presence ofβ′-formin glyceryl behenate nanoparticles after incorporation into an oil/water cream.

5. Conclusion

DSC and X-ray/neutron diffraction and scattering techniquesare indispensable tools for LNP characterization and providemany possibilities to derive information on the properties of thedispersed particles. Although the use of these techniques isstraightforward in some cases, other applications in the field ofLNP research require profound knowledge on the physico-chemical properties of the dispersions as well as on thespecificities of the respective instrumental technique. Detailedknowledge on the analytical procedures and their potential does,however, also allow to derive much more information on thesample properties than commonly obtained today. Althoughthey do require a somewhat larger analytical effort particularlythe scattering techniques which bear a large potential for theinvestigation of LNP dispersions that remains to be fullyexplored.

Acknowledgements

The authors would like to thank Judith Kuntsche for drawingsome of the figures and Andreas Illing for providing SAXS andSANS images.

Reference

[1] P. Kinnunen, J.-M. Alakoskela, P. Laggner, Phase behavior of liposomes,Methods Enzymol. 367 (2003) 129–147.

[2] R.L. Biltonen, D. Lichtenberg, The use of differential scanningcalorimetry as a tool to characterize liposome preparations, Chem.Phys. Lipids 64 (1993) 129–142.

[3] K.M.G. Taylor, R.M. Morris, Thermal analysis of phase transitionbehavior in liposomes, Thermochim. Acta 248 (1995) 289–301.

[4] J. Gustafsson, H. Ljusberg-Wahren, K. Almgren, M. Larsson, Submicronparticles of reversed lipid phases in water stabilized by a nonionicamphiphilic polymer, Langmuir 13 (1997) 6964–6971.

[5] G. Wörle, B. Siekmann, M.H.J. Koch, H. Bunjes, Transformation ofvesicular into cubic nanoparticles by autoclaving of aqueous monoolein/poloxamer dispersions, Eur. J. Pharm. Sci. 27 (2006) 44–53.

[6] P. Walstra, E.C.H. van Beresteyn, Crystallization of milk fat in theemulsified state, Neth. Milk Dairy J. 29 (1975) 35–65.

[7] W. Kloek, P. Walstra, T. van Vliet, Nucleation kinetics of emulsifiedtriglyceride mixtures, J. Am. Oil Chem. Soc. 77 (2000) 643–652.

[8] T. Awad, K. Sato, Effects of hydrophobic additives on crystallizationbehavior of palm mid fraction in oil-in-water emulsion, J. Am. Oil Chem.Soc. 78 (2001) 837–842.

[9] J. Coupland, Crystallization in emulsions, Curr. Opin. Colloid InterfaceSci. 7 (2002) 445–450.

[10] D. Kalnin, O. Schafer, H. Amenitsch, M. Ollivon, Fat crystallization inemulsion: influence of emulsifier concentration on triacylglycerol crystalgrowth and polymorphism, Cryst. Growth Des. 4 (2004) 1283–1293.

[11] R.J. Deckelbaum, G.G. Shipley, D.M. Small, Structure and interactions oflipids in human plasma low density lipoproteins, J. Biol. Chem. 252(1977) 744–754.

[12] D.A. Waugh, D.M. Small, Rapid method for determining cholesteryl estertransitions of apoB-containing lipoproteins, J. Lipid Res. 23 (1982) 201–204.

[13] P. Laggner, X-ray and neutron small-angle scattering on plasmalipoproteins, NATO ASI Ser., C 451 (1995) 371–386.

[14] R. Prassl, B. Schuster, P.M. Abuja, M. Zechner, G.M. Kostner, P.Laggner, A Comparison of structure and thermal behavior in humanplasma lipoprotein(a) and low-density lipoprotein. Calorimetry andsmall-angle X-ray scattering, Biochemistry 34 (1995) 3795–3801.

[15] M. von Laue, Röntgenstrahlinterferenzen, Akademische Verlagsge-sellschaft, Frankfurt am Main, 1960.

[16] P. Luger, Modern X-Ray Analysis on Single Crystals, de Gruyter, Berlin,1980.

[17] H.P. Klug, L.E. Alexander (Eds.), X-Ray Diffraction Procedures, 3rdEdition, John Wiley and Sons, London, 1962.

[18] A. Guinier, G. Fournet, Small-Angle Scattering of X-Rays, John Wiley &Sons, New York, 1955.

[19] G.L. Squires, Thermal Neutron Scattering, Cambridge University Press,Cambridge, 1978.

[20] S. Haussuhl, Kristallgeometrie, 2nd Edition, VCH Verlagsgesellschaft,Weinheim, 1993.

[21] C. Washington, M.J. Lawrence, D. Barlow, Neutron scattering inpharmaceutical sciences, in: B.J. Gabrys (Ed.), Applications of NeutronScattering to Soft Condensed Matter, Gordon and Breach SciencePublishers, Amsterdam, 2000, pp. 325–356.

[22] M. Bée, Quasielastic Neutron Scattering, Adam Hilger, Bristol, 1988.[23] K. Jores, A. Haberland, S. Wartewig, K. Mäder, W. Mehnert, Solid lipid

nanoparticles (SLN) and oil-loaded SLN studied by spectrofluorometryand Raman spectroscopy, Pharm. Res. 22 (2005) 1887–1897.

[24] K. Jores, W. Mehnert, M. Drechsler, H. Bunjes, C. Johann, K. Mäder,Investigations on the structure of solid lipid nanoparticles (SLN) and oil-loaded solid lipid nanoparticles by photon correlation spectroscopy, field-flow fractionation and transmission electron microscopy, J. Control.Release 95 (2004) 217–227.

[25] L. de Broglie, Licht und Materie, Goverts, Hamburg, 1939.[26] H. Rauch, W. Waschkowski, Neutron scattering lengths, in: H. Schopper

(Ed.), Landolt-Börnstein, Vol. I/16A of New Series, Springer, Berlin,2000, pp. 1–29, Ch. 6.

[27] S.K. Sinha, M. Tolan, A. Gibaud, Effects of partial coherence on thescattering of X-rays by matter, Phys. Rev., B 57 (1998) 2740–2758.

[28] D.T. Cromer, J.T. Waber, Atomic Scattering Factors for X-Rays,International Tables for X-Ray Crystallography, vol. 4, Kynoch Press,Birmingham, 1974, pp. 71–147.

[29] T. Unruh, H. Bunjes, K. Westesen, M.H.J. Koch, Observation of size-dependent melting in lipid nanoparticles, J. Phys. Chem., B 103 (1999)10373–10377.


[30] H. Bunjes, M.H.J. Koch, K. Westesen, Effect of particle size on colloidalsolid triglycerides, Langmuir 16 (2000) 5234–5241.

[31] H. Bunjes, M.H.J. Koch, K. Westesen, Influence of emulsifiers on thecrystallization of solid lipid nanoparticles, J. Pharm. Sci. 92 (2003)1509–1520.

[32] A. Illing, T. Unruh, Investigations on the flow behavior of dispersions ofsolid triglyceride nanoparticles, Int. J. Pharm. 284 (2004) 123–131.

[33] T. Unruh, H. Bunjes, K. Westesen, M.H.J. Koch, Investigations on themelting behaviour of triglyceride nanoparticles, Colloid Polym. Sci. 279(2001) 398–403.

[34] A. van Langevelde, K. van Malssen, F. Hollander, R. Peschar, H. Schenk,Structure of mono-acid even-numbered β-triacylglycerols, Acta Crystallogr.,B 55 (1999) 114.

[35] D.P. Tielemann, H.J.C. Berendsen, Molecular dynamics simulations of afully hydrated dipalmitoylphosphatidylcholine bilayer with differentmacroscopic boundary conditions and parameters, J. Chem. Phys. 105(1996) 4871–4880.

[36] H. Bunjes, F. Steininger, W. Richter, Visualizing the structure oftriglyceride nanoparticles in different crystal modifications, Langmuir23 (2007) 4005–4011.

[37] T. Unruh, Interpretation of SAXS patterns of crystalline triglyceridenanoparticles in dispersion, J. Appl. Cryst. (submitted for publication).

[38] T. Unruh, K. Westesen, P. Bösecke, P. Lindner, M.H.J. Koch, Self-assembly of triglyceride nanocrystals in suspension, Langmuir 18 (2002)1796–1800.

[39] A. Illing, T. Unruh, M.H.J. Koch, Investigation on particle self-assemblyin solid lipid based colloidal drug carrier systems, Pharm. Res. 21 (2004)592–597.

[40] T. Unruh, A. Illing, Investigation on the influence of shear on a preferredorientation of triglyceride nanocrystals in suspension, Tech. Rep., ILLExperimental Report, 2003.

[41] A. Illing, T. Unruh, M.H.J. Koch, Investigation on the incorporation ofdrugs into triglyceride nanosuspensions, Tech. Rep., EMBLExperimentalReport, 2003.

[42] P. Bergese, I. Colombo, D. Gervasoni, L.E. Depero, Melting ofnanostructured drugs embedded into a polymeric matrix, J. Phys.Chem. B 108 (2004) 15488–15493.

[43] H. Bunjes, M. Drechsler, M.H.J. Koch, K. Westesen, Incorporation of themodel drug ubidecarenone into solid lipid nanoparticles, Pharm. Res. 18(2001) 287–293.

[44] K. Jores, W. Mehnert, K. Mäder, Physicochemical investigations on solidlipid nanoparticles and on oil-loaded solid lipid nanoparticles: a nuclearmagnetic resonance and electron spin resonance study, Pharm. Res. 20(2003) 1274–1283.

[45] V. Jenning, S.H. Gohla, Encapsulation of retinoids in solid lipidnanoparticles (SLN), J. Microencapsul. 18 (2001) 149–158.

[46] K. Westesen, H. Bunjes, M.H.J. Koch, Physicochemical character-ization of lipid nanoparticles and evaluation of their drug loadingcapacity and sustained release potential, J. Control. Release 48 (1997)223–236.

[47] H. Bunjes, K. Westesen, M.H.J. Koch, Crystallization tendency andpolymorphic transitions in triglyceride nanoparticles, Int. J. Pharm. 129(1996) 159–173.

[48] V. Jenning, M. Schäfer-Korting, S.H. Gohla, Vitamin A loaded solid lipidnanoparticles for topical use: drug release properties, J. Control. Release66 (2000) 115–126.

[49] M. Higami, S. Ueno, T. Segawa, K. Iwanami, K. Sato, Simultaneoussynchrotron radiation X-ray diffraction–DSC analysis of melting andcrystallization behavior of trilauroylglycerol in nanoparticles of oil-in-water emulsion, J. Am. Oil Chem. Soc. 80 (2003) 731–739.

[50] G. Höhne, W. Hemminger, H.-J. Flammersheim, Differential ScanningCalorimetry, Springer, Heidelberg, 2003.

[51] B. Wunderlich, Thermal Analysis, Academic Press, New York, 1990.[52] J.L. Ford, P. Timmins, Pharmaceutical Thermal Analysis—Techniques

and Applications, Ellis Horwood, Chichester, 1989.[53] D. Giron, Thermal analysis of drugs and drug products, in: J. Swarbrick,

J.C. Boylan (Eds.), Encyclopedia of Pharmaceutical Technology, 2ndEdition, Marcel Dekker, New York, 2002, pp. 2766–2793.

[54] S.D. Clas, C.R. Dalton, B. Hancock, Calorimetry in pharmaceuticalresearch and development, in: J. Swarbrick, J.C. Boylan (Eds.),Encyclopedia of Pharmaceutical Technology, 2nd Edition, MarcelDekker, New York, 2002, pp. 289–301.

[55] J.P. Dumas, Y. Zeraouli, M. Strub, Heat transfer inside emulsionsdetermination of the DSC thermograms part 1. Crystallization of theundercooled droplets, Thermochim. Acta 236 (1994) 227–237.

[56] G. Höhne, W. Hemminger, H.-J. Flammersheim, Differential ScanningCalorimetry, Springer, Heidelberg, 2003, pp. 72–73.

[57] A. zur Mühlen, C. Schwarz, W. Mehnert, Solid lipid nanoparticles (SLN)for controlled drug delivery— drug release and release mechanism, Eur.J. Pharm. Biopharm. 45 (1998) 149–155.

[58] C. Schwarz, W. Mehnert, Freeze-drying of drug-free and drug-loadedsolid lipid nanoparticles (SLN), Int. J. Pharm. 157 (1997) 171–179.

[59] C. Schwarz, W. Mehnert, Solid lipid nanoparticles (SLN) for controlleddrug delivery II. Drug incorporation and physicochemical characteriza-tion, J. Microencapsul. 16 (1999) 205–213.

[60] B. Siekmann, K. Westesen, Thermoanalysis of the recrystallizationprocess of melt-homogenized glyceride nanoparticles, Colloids Surf. B:Biointerfaces 3 (1994) 159–175.

[61] H. Bunjes, B. Siekmann, K. Westesen, Emulsions of supercooled melts—a novel drug delivery system, in: S. Benita (Ed.), Submicron Emulsions inDrug Targeting and Delivery, Harwood Academic Publishers, Amsterdam,1998, pp. 175–218.

[62] K. Westesen, H. Bunjes, Do nanoparticles prepared from lipids solid atroom temperature always possess a solid lipid matrix? Int. J. Pharm. 115(1995) 129–131.

[63] S. Lim, C.K. Kim, Formulation parameters determining the physico-chemical characteristics of solid lipid nanoparticles loaded with all-transretinoic acid, Int. J. Pharm. 243 (2002) 135–146.

[64] J. Kuntsche, K. Westesen, M. Drechsler, M.H.J. Koch, H. Bunjes,Supercooled smectic nanoparticles— a potential novel carrier system forpoorly water soluble drugs, Pharm. Res. 21 (2004) 1836–1845.

[65] M.J.W. Povey, Crystallization of oil-in-water emulsions, in: N. Garti, K.Sato (Eds.), Crystallization Processes in Fats and Lipid Systems, MarcelDekker, New York, 2001, pp. 177–209.

[66] M.A. Schubert, B.C. Schicke, C.C. Muller-Goymann, Thermal analysisof the crystallization and melting behavior of lipid matrices and lipidnanoparticles containing high amounts of lecithin, Int. J. Pharm. 298(2005) 242–254.

[67] T. de Vringer, H.A.G. de Ronde, Preparation and structure of a water-in-oil cream containing lipid nanoparticles, J. Pharm. Sci. 84 (1995)466–472.

[68] V. Jenning, A.F. Thünemann, S.H. Gohla, Characterisation of a novelsolid lipid nanoparticle carrier system based on binary mixtures of liquidand solid lipids, Int. J. Pharm. 199 (2000) 167–177.

[69] E. Zimmermann, E.B. Souto, R.H. Müller, Physicochemical investigationson the structure of drug-free and drug-loaded solid lipid nanoparticles (SLN)by means of DSC and 1H NMR, Pharmazie 60 (2005) 508–513.

[70] B. Siekmann, K. Westesen, Preparation and physicochemical character-ization of aqueous dispersions of coenzyme Q10 nanoparticles, Pharm.Res. 12 (1995) 201–208.

[71] R.J. Hunter, Foundations of Colloid Science, vol. I, Clarendon Press,Oxford, 1993, pp. 267–269.

[72] W. Skoda, L.L. Hoekstra, T.C. van Soest, P. Bennema, M. van denTempel, Structure and morphology of β-crystals of glyceryl tristearate,Kolloid-Z. Z. Polym. 219 (1967) 149–156.

[73] D.M. Small, General properties of lipids conferred by the aliphatic chain,Handbook of Lipid Research, vol. 4, Plenum Press, New York, 1986,pp. 21–41, Ch. 2.

[74] K. Sato, Polymorphism of pure triacylglycerols and natural fats, Adv.Lipid Res. 2 (1996) 213–268.

[75] J.W. Hagemann, Thermal behavior and polymorphism of acylglycerides,in: N. Garti, K. Sato (Eds.), Crystallization and Polymorphism of Fats andFatty Acids, Marcel Dekker, New York, 1988, pp. 9–95.

[76] L. Hernqvist, Crystal structures of fats and fatty acids, in: N. Garti, K.Sato (Eds.), Crystallization and Polymorphism of Fats and Fatty Acids,Marcel Dekker, New York, 1988, pp. 97–137.


[77] K. Sato, S. Ueno, Molecular interactions and phase behavior of poly-morphic fats, in: N. Garti, K. Sato (Eds.), Crystallization Processes in Fatsand Lipid Systems, Marcel Dekker, New York, 2001, pp. 177–209.

[78] F. Kaneko, Polymorphism and phase transitions of fatty acids andacylglycerols, in: N. Garti, K. Sato (Eds.), Crystallization Processes inFats and Lipid Systems, Marcel Dekker, New York, 2001, pp. 53–97.

[79] K. Sato, Crystallization of fats and fatty acids, in: N. Garti, K. Sato (Eds.),Crystallization and Polymorphism of Fats and Fatty Acids, MarcelDekker, New York, 1988, pp. 227–263.

[80] K. Sato, M. Goto, J. Yano, K. Honda, D.R. Kodali, D.M. Small, Atomicresolution structure analysis of β′ polymorph crystal of a triacylglycerol:1,2-dipalmitoyl-3-myristoyl-sn-glycerol, J. Lipid Res. 42 (2001)338–345.

[81] R.E. Timms, Phase behavior of fats and their mixtures, Prog. Lipid Res.23 (1984) 1–38.

[82] K. Westesen, B. Siekmann, M.H.J. Koch, Investigations on the physicalstate of lipid nanoparticles by synchrotron radiation X-ray diffraction, Int.J. Pharm. 93 (1993) 189–199.

[83] H. Bunjes, M.H.J. Koch, Saturated phospholipids promote crystallizationbut slow down polymorphic transitions in triglyceride nanoparticles,J. Control. Release 107 (2005) 229–243.

[84] M. Kellens,W.Meussen, R. Gehrke, H. Reynaers, Synchrotron investigationsof the polymorphic transitions of saturated monacid triglycerides. Part 1:tripalmitin and tristearin, Chem. Phys. Lipids 58 (1991) 131.

[85] M.A. Schubert, C.C. Müller-Goymann, Characterisation of surface-modified solid lipid nanoparticles (SLN): influence of lecithin andnonionic emulsifier, Eur. J. Pharm. Biopharm. 61 (2005) 77–86.

[86] C.W. Hoerr, Morphology of fats, oils and shortenings, J. Am. Oil Chem.Soc. 37 (1960) 539–546.

[87] K. Thoma, P. Serno, D. Precht, Röntgendiffraktometrischer Nachweis derPolymorphie von Hartfett, Pharm. Ind. 45 (1983) 420–425.

[88] J.H. Whittam, H.L. Rosano, Physical aging of even saturated monoacidtriglycerides, J. Am. Oil Chem. Soc. 52 (1975) 128–133.

[89] D.M. Small, Substituted aliphatic hydrocarbons: alcohols and acids,Handbook of Lipid Research, vol. 4, Plenum Press, New York, 1986,pp. 233–284.

[90] D. Aquilano, R. Cavalli, M.R. Gasco, Solid lipospheres from hotmicroemulsions in the presence of different concentrations of cosurfactant:the crystallization of stearic acid polymorphs, Thermochim. Acta 230 (1993)29–37.

[91] R. Cavalli, D. Aquilano, M.E. Carlotti, M.R. Gasco, Study by X-raypowder diffraction and differential scanning calorimetry of two modeldrugs, phenothiazine and nifedipine, incorporated into lipid nanoparti-cles, Eur. J. Pharm. Biopharm. 41 (1995) 329–333.

[92] R. Cavalli, O. Caputo, E. Ugazio, M.R. Gasco, The effect of butanol andpentanol isomers on the crystallization of stearic acid polymorphs in solidlipid nanoparticles, Acta Technol. Legis Medicam. 7 (1996) 101–111.

[93] R. Cavalli, O. Caputo, E. Marengo, F. Pattarino, M.R. Gasco, The effectof the components of microemulsions on both size and crystallinestructure of solid lipid nanoparticles (SLN) containing a series of modelmolecules, Pharmazie 53 (1998) 392–396.

[94] N. Garti, S. Sarig, E. Wellner, Determination of the compositions of mixturesof fatty acid polymorphs by DTA, Thermochim. Acta 37 (1980) 131–134.

[95] K. Westesen, M. Drechsler, H. Bunjes, Colloidal dispersions based onsolid lipids, in: E. Dickinson, R. Miller (Eds.), Food Colloids:Fundamentals of Formulation, Royal Society of Chemistry, Cambridge,2001, pp. 103–115.

[96] C. Freitas, R.H. Müller, Correlation between long-term stability of solidlipid nanoparticles (SLN) and crystallinity of the lipid phase, Eur. J.Pharm. Biopharm. 47 (1999) 125–132.

[97] C. Olbrich, O. Kayser, R.H. Müller, Lipase degradation of Dynasan 114and 116 solid lipid nanoparticles (SLN) — effect of surfactants, storagetime and crystallinity, Int. J. Pharm. 237 (2002) 119–128.

[98] S.A. Wissing, R.H. Müller, The influence of the crystallinity of lipidnanoparticles on their occlusive properties, Int. J. Pharm. 242 (2002) 377–379.

[99] M. Üner, S.A. Wissing, G. Yener, R.H. Müller, Solid lipid nanoparticles(SLN) and nanostructured lipid carriers (NLC) for application of ascorbylpalmitate, Pharmazie 60 (2005) 577–582.

[100] J. Kuntsche, M.H.J. Koch, M. Drechsler, H. Bunjes, Crystallizationbehavior of supercooled smectic cholesteryl myristate nanoparticlescontaining phospholipids as stabilizers, Colloids Surf. B: Biointerfaces44 (2005) 25–35.

[101] A. Lippacher, R.H. Müller, K. Mäder, Investigation on the viscoelasticproperties of lipid based colloidal drug carriers, Int. J. Pharm. 196 (2000)227–230.

[102] H. Bunjes, M.H.J. Koch, K. Westesen, Effects of surfactants on thecrystallization and polymorphism of lipid nanoparticles, Prog. Colloid &Polym. Sci. 121 (2002) 7–10.

[103] T. Katsuragi, N. Kaneko, K. Sato, Effects of addition of hydrophobic sucrosefatty acid oligoesters on crystallization rates of n-hexadecane in oil-in-wateremulsions, Colloids Surf. B: Biointerfaces 20 (2001) 229–237.

[104] T. Awad, Y. Hamada, K. Sato, Effects of addition of diacylglycerols on fatcrystallization in oil-in-water-emulsions, Eur. J. Lipid Sci. Technol. 103(2001) 735–741.

[105] T. Awad, K. Sato, Acceleration of crystallization of palm kernel oil in oil-in-water emulsion by hydrophobic emulsifier additives, Colloids Surf., BBiointerfaces 25 (2002) 45–53.

[106] T.S. Awad, K. Sato, Fat crystallization in o/w emulsions controlled byhydrophobic emulsifier additives, in: A.G. Marangoni, S.S. Narine (Eds.),Physical Properties of Lipids, Marcel Dekker, New York, 2002, pp. 37–62.

[107] H. Bunjes, J. Kuntsche, Supercooled smectic nanoparticles, in: D.Thassu, M. Deleers, Y. Pathak (Eds.), Nanoparticulate Drug DeliverySystems, Informa Healthcare, New York, 2007, pp. 129–140.

[108] V. Jenning, K. Mäder, S.H. Gohla, Solid lipid nanoparticles (SLN) basedon binary mixtures of liquid and solid lipids: a 1H-NMR study, Int. J.Pharm. 205 (2000) 15–21.

[109] F. Castelli, C. Puglia, M.G. Sampietro, L. Rizza, F. Bonina, Character-ization of indomethacin-loaded solid lipid nanoparticles by differentialscanning calorimetry, Int. J. Pharm. 304 (2005) 231–238.

[110] M. Garcia-Fuentes, M.J. Alonso, D. Torres, Design and characterizationof a new drug nanocarrier made from solid–liquid lipid mixtures,J. Colloid Interface Sci. 285 (2005) 590–598.

[111] R.H. Müller, M. Radtke, S.A. Wissing, Nanostructured lipid matrices forimproved microencapsulation of drugs, Int. J. Pharm. 242 (2002) 121–128.

[112] F.-Q. Hu, S.-P. Jiang, Y.-Z. Du, H. Yuan, Y.Q. Ye, S. Zeng, Preparationand characterization of stearic acid nanostructured lipid carriers bysolvent diffusion method in an aqueous system, Colloids Surf. B:Biointerfaces 45 (2005) 167–173.

[113] J.B. Rossell, Phase diagrams of triglyceride systems, in: R. Paoletti, D.Kritchevsky (Eds.), Advances in Lipid Research, vol. V, Academic Press,New York, 1967, pp. 353–408.

[114] M. Takeuchi, S. Ueno, K. Sato, Synchrotron radiation SAXS/WAXS studyof polymorph-dependent phase behavior of binary mixtures of saturatedmonoacid triacylglycerols, Cryst. Growth Des. 3 (2003) 369–374.

[115] G. Höhne, W. Hemminger, H.-J. Flammersheim, Differential ScanningCalorimetry, Springer, Heidelberg, 2003, pp. 230–232.

[116] J.L. Ford, P. Timmins, Pharmaceutical Thermal Analysis—Techniquesand Applications, Ellis Horwood, Chichester, 1989, pp. 47–61.

[117] B. Wunderlich, Thermal Analysis, Academic Press, New York, 1990,pp. 171–177.

[118] H. Yoshino, M. Kobayashi, M. Samejima, Influence of the liquid phasecoexisting in fatty suppository bases on the polymorphic transition rate,Chem. Pharm. Bull. 30 (1982) 2941–2950.

[119] I. Friedrich, C.C. Müller-Goymann, Characterization of solidified reversemicellar solutions (SRMS) and production development of SRMS-basednanosuspensions, Eur. J. Pharm. Biopharm. 56 (2003) 111–119.

[120] M.A. Schubert, M. Harms, C.C. Müller-Goymann, Structural investiga-tions on lipid nanoparticles containing high amounts of lecithin, Eur. J.Pharm. Sci. 27 (2006) 226–236.

[121] R.A. Young (Ed.), The Rietveld Method, Oxford University Press, 1993.[122] V. Venkateswarlu, K. Manjunath, Preparation, characterization and in

vitro release kinetics of clozapine solid lipid nanoparticles, J. Control.Release 95 (2004) 627–638.

[123] R. Cavalli, O. Caputo, M.E. Carlotti, M. Trotta, C. Scarnecchia, M.R.Gasco, Sterilization and freeze-drying of drug-free and drug-loaded solidlipid nanoparticles, Int. J. Pharm. 148 (1997) 47–54.


[124] R. Cavalli, E. Peira, O. Caputo, M.R. Gasco, Solid lipid nanoparticles ascarriers of hydrocortisone and progesterone complexes with β-cyclodex-trins, Int. J. Pharm. 182 (1999) 59–69.

[125] S.A. Wissing, R.H. Müller, A novel sunscreen system based ontocopherol acetate incorporated into solid lipid nanoparticles, Int. J.Cosmet. Sci. 23 (2001) 233–243.

[126] S.C. Yang, J.B. Zhu, Preparation and characterization of camptothecinsolid lipid nanoparticles, Drug Devel. Ind. Pharm. 28 (2002) 265–274.

[127] H. Bunjes, Influence of different factors on the structure and properties ofnanoparticles from solid triglycerides, Ph. D. thesis, Friedrich-Schiller-Universität Jena, Jena (1998).

[128] S.A. Wissing, R.H. Müller, The influence of solid lipid nanoparticleson skin hydration and viscoelasticity — in vivo study, Eur. J. Pharm.Biopharm. 56 (2003) 67–72.

Documents

Characterization of lipid nanoparticles by differential ...prescientdrugdelivery.com/wp-content/uploads/2012/... · X-ray interferences (Debye–Waller factor) and corrections specific