Previous PageNext Page

Chapter 2 - Materials and Methods

2.1 Materials

2.1.1 Emulsifiers

Proprietary name




Lipoid E75®

Lipoid KG, Ludwigshafen, Germany


72.6% PC, 13.5% PE,
2.6% LPC, 2.3% SPM

Lipoid E80®

Lipoid KG, Ludwigshafen, Germany


77.7% PC, 7.8% PE,
2.5% LPC, 3.0% SPM

Lipoid EPC®

Lipoid KG, Ludwigshafen, Germany

T 12085-1

98.0% PC,
< 0.2% LPC

Lipoid ELPC®

Lipoid KG, Ludwigshafen, Germany

T 22014

0.3% PC,
99.0% LPC


Avanti Polar Lipids, Alabaster, AL,USA

(used for 31P-NMR experiments)

> 99 % LPC

Egg-Phosphatidyl-ethanolamine Type III P7943

Sigma Chemicals GmbH, Deisenhofen, Germany


> 99 % PE

Egg-Lysophospha-tidylcholine Type I L4129

Sigma Chemicals GmbH, Deisenhofen, Germany


> 99% LPC


Lipoid KG, Ludwigshafen, Germany



Lutrol (Pluronic) F 68®

BASF AG, Ludwigshafen, Germany


Poloxamer 188

Texapon K 12®-96C

Henkel KGaA, Duesseldorf, Germany


Sodium lauryl-sulphate

Sodium Oleate

Aldrich Chemicals, Deisenhofen, Germany


> 98% Sodium cis-9-

(PC = phosphatidylcholine; PE = phosphatidylethanolamine; LPC = lyso-PC; SPM = sphingomyelin)

All lecithins and sodium oleate were stored under N2 and at -20°C. Pluronic® and Texapon® were stored tightly closed at room temperature. Composition is given according to the manufacturer’s declaration.

2.1.2 Oils and waxes

Proprietary name




Soybean Oil S7381

Sigma Chemicals GmbH, Deisenhofen, Germany


(Complied with USP and BP requirements)

Paraffin viscous, USP, Ph.Eur.

Merck KGaA, Darmstadt, Germany


: 110-230 mPas

Paraffin liquid, USP, Ph.Eur.

BP Oiltech, Hamburg, Germany

Enerpar M002

: 26.2 mPas

Cutina CP®

Henkel KgaA, Duesseldorf, Germany


Hexadecanoic acid hexadecyl ester

Soybean oil was stored away from light under N2 at room temperature. Cutina® and paraffins were stored tightly closed at room temperature. Composition is given according to the manufacturer’s declaration.

2.1.3 Additives

Proprietary name




Glycerol, Ph.Eur.

Carl Roth GmbH&Co, Karlsruhe, Germany


Glycerol, double-distilled, 98%

Sodium hydroxide purum, p.a. grade

Fluka Chemie AG, Deisenhofen, Germany

347308/1 396

> 98% NaOH

Palmitic acid

Sigma Chemicals GmbH, Deisenhofen, Germany


> 99% Hexadecanoic

Oleic acid O3879

Sigma Chemicals GmbH, Deisenhofen, Germany


Ca. 99%
Cis-9-Octadecenoic acid

Fatty acids were stored under N2 at -20°C. Glycerol was stored tightly closed under N2 at room temperature. Sodium hydroxide was stored tightly closed; solutions were prepared freshly and stored at room temperature. Composition is given according to the manufacturer’s declaration.

2.1.4 Parenteral Fat Emulsions

Proprietary name




10 / 20 / 30

Pharmacia&Upjohn GmbH, Erlangen, Germany

68460-51 & 87535-51 (10%)
70178-51 & 85375-51 (20%)
88555-71 & 87537-51 (30%)

3/97 & 1/99
5/97 & 8/98
3/99 & 1/99

10% N / 20% N

B.Braun Melsungen AG, Melsungen, Germany

5471A81 (10%)
6071A81 (20%)

10/97 (10%)
1/98 (20%)

Lipofundin MCT® (aka Medialipide®, Vasolipid®)
10% / 20 %

B.Braun Melsungen AG, Melsungen, Germany

6023A81 (10%)
6031A81 (20%)

12/97 (10%)
12/97 (20%)

LCT 10 PLR / 20

Fresenius AG, Bad Homburg v.d.H., Germany

FG 1013P (10%PLR)
FK 1005 (20%)

1/97 (10%PLR)
10/97 (20%)

Original emulsion bottles were stored away from light at room temperature.

2.1.5 Reagents

Proprietary name




Copper sulphate, anhydrous

Fluka Chemie AG, Deisenhofen, Germany



Potassium chloride, extra pure

Merck KGaA, Darmstadt, Germany

023 TA 839135


Sulphuric acid, purum p.a. grade

Fluka Chemie AG, Deisenhofen, Germany

3523 90/1 1295

95-97% H2SO4

Phosphoric acid, extra pure

Merck KGaA, Darmstadt, Germany

810 K040 186 63

85% ortho-H3PO4-solution

Hydrochloric acid, purum p.a. grade

Fluka Chemie AG, Deisenhofen, Germany

3585 33/1 34596

37% HCl-solution

2.1.6 Solvents

Proprietary name




2-Propanol, gradient grade (LiChrosolv®)

Merck KgaA, Darmstadt, Germany

I 732340 723

> 99.8% 2-Propanol

n-Hexane, HPLC grade (LiChrosolv®)

Merck KGaA, Darmstadt, Germany

I 757591 740

> 97% n-Hexane

Water for HPLC


Freshly prepared in a Destamat Bi 18 T, (Heraeus, Germany)

Double-distilled water 0.22 µm-vacuum-filtered

Glycerol/Water for Particle Sizing Dilution


Kept in stock after

2.25% (w/w) Glycerol/Water 0.22µm-filtered

Water for Langmuir Film Balance


Freshly prepared in a Seralpur DELTA UV, (Seral, Germany)

Ultra-pure water 0.2 µm-filtered (Supor DCF filter)

Methanol, p.a. ACS grade (Rotipuran®)

Carl Roth GmbH&Co, Karlsruhe, Germany


> 99.8% Methanol

Chloroform, p.a. (Rotipuran®)

Carl Roth GmbH&Co, Karlsruhe, Germany


> 99% Trichloromethane

Chloroform, spectroscopy grade (Uvasol®)

Merck KGaA, Darmstadt, Germany

I 375547

> 98% purity, 0.0005% residue on evaporation

Ammonia solution,
extra pure

Merck KGaA, Darmstadt, Germany


32% NH3-solution

2.2 Methods

2.2.1 Investigation of Emulsifying Properties Measurement of Droplet Coalescence Times with Coalescence Cell

A droplet-to-planar-surface coalescence model was used to study survival times of oil droplets released to coalesce at a soybean oil-water interface as a function of addition of phospholipids. The coalescence cell consisted of a double-walled glass apparatus connected to a Haake FE 2 thermostatting device (Fig. 12). A micrometer-controlled CR-700-200 type Hamilton® syringe equipped with a customised, curved needle (end-to-end length: 10 cm, : 0.4 mm) with repeated-action function was fitted through an acrylic glass lid, which covered the cell and also kept the syringe fixed in position. To adapt droplet ejection speed to the viscosity of the oil phase, a different spring was used in the syringe. Thus, for each ejection oil was drawn from the upper phase and released, yielding uniform, single droplets of 20 µl. Before each experiment the syringe and the glass apparatus were thoroughly rinsed with 2-propanol, double-distilled water followed by sulfurochromic acid and again double-distilled water. The oil-phase (10 g soya oil) containing the emulsifier was layered on the water phase (90 g) and 100 droplets counted. Each experiment was repeated at least once. Determination of Emulsifier Film Compressibility with Langmuir Film Balance

Recording of film pressure-molecular area (-A) isotherms was carried out with an FW-2 type Langmuir film balance (Lauda GmbH&Co KG, Lauda-Koenigshofen, Germany) connected to a thermostatting bath (RC20CP, Lauda GmbH&Co KG) which allowed measurements at 0°C - 55°C 0.1°C. Fig. 13 shows the principle of film pressure measurement by a Langmuir film balance of insoluble films from amphiphilic molecules at the air-liquid interface.

Fig. 13 Principle of film pressure measurement with a Langmuir film balance

Monomolecular Langmuir-Blodgett films of amphiphilic compounds at air-water interfaces, are prepared by dissolving the amphiphile into a suitable solvent which by rapid evaporation leaves a two-dimensional film of the amphiphile when the solution is spread across the surface of a second liquid, which must be immiscible with the solution. Typical solvents used for preparation of such films are e.g. hexane or chloroform [Gaines, 1966], which are also suitable for the case of lecithin. Solvents should be of high purity, since any remnants after evaporation capable of film-formation would lead to erroneous results. Best results were found for 2 mM chloroformic solutions, which were spread at various points of the subphase and allowed to evaporate. Film pressure-area dependence can be measured with either a Langmuir film balance or a Wilhelmy-plate [Gaines, 1966]. Both methods use either Teflon or glass troughs containing the subphase, a moveable barrier which allows compression (or expansion) of the film floating on top of the subphase, and an assembly to record the forces transduced by the film [Shaw, 1983 / Ulman, 1991]. In the case of the Langmuir film balance used in this study, a fixed barrier float is employed to continuously record the film pressure [mN/m] of the monolayer against the barrier when being compressed by the moveable barrier.

Calibration of area and pressure was carried out automatically by the film balance before each measurement. Monolayers were spread on the cleaned subphase consisting of high-purity water (see Materials), which was verified by measurement of two consecutive zero-isotherms before each run. Each measurement was carried out in triplicate using 25 µl of solution metered by a Hamilton syringe. Before starting compression, chloroform was allowed to evaporate for 10 min. The complete isotherm measurement was preset to take 40 min in order to allow careful compression of the film (approx. 225 mm2/min). At the end of each measurement the compressed film was drawn from the surface by means of six aspiration nozzles located at the fixed barrier and connected to a vacuum pump, and highly purified water was refilled from behind the moveable barrier.

2.2.2 Preparation of Emulsions, Solid Lipid Nanoparticles (SLN) and Liposomes

For preparation of pre-emulsions, phospholipid pre-dispersions and solutions two JKAMAG RET type magnetic stirrers/heaters were employed equipped with a JKA-TRON® ETS-D thermostatting control unit (JKA, Staufen, Germany) to maintain preset temperature values. Stirring speed was approx. 700 rpm. Unless stated otherwise, lecithin was dispersed, or Pluronic® or sodium oleate were dissolved in double-distilled water which had been heated to 60°C, then glycerol was added. Subsequently, in the case of emulsions and SLNs, the oil or melted wax heated to 60°C was slowly incorporated into the lecithin dispersion. After about 15 min of stirring, the final weight was adjusted with double-distilled water. A typical emulsion formulation consisted of:

1) Oil phase (e.g. soya oil) 5 / 10 / 20 or 30 g
2) Emulsifier (e.g. lecithin) 0.6 or 1.2 g
3) Sodium oleate 0.03 g (only where indicated)
4) Glycerol 2.25 g
5) Double-distilled water to 100.0 g

Liposomal dispersions were of the same formulation by omitting addition of the oil. Customised Evacuable Ultra-Turrax® Mixer

Increase in state of dispersion of the pre-emulsions or liposome dispersions was achieved using a customised high-speed stirrer. An JKA Ultra-Turrax® TP 18/10 stirrer with an 18K type vacuum homogenising rotor coupled to a Thyristor JKA TR50 for speed control (JKA, Staufen, Germany) was fitted into an evacuable steel jacket with a glass window for visual control (Fig. 15). Approx. 5000 rpm was applied for approx. 5 min to homogenise the samples in a glass beaker placed inside the jacket. High-speed stirring always leads to incorporation of air bubbles which deplete the emulsion of emulsifier and, furthermore, necessitates additional nitrogen flushing to prevent fast oxidation of unsaturated components by incorporated oxygen. By evacuating the jacket by a RD-2 Vacuubrand® vacuum pump (R. Brand, Wertheim, Germany) to approx. 310-2 mbar, incorporation of gas during stirring could be avoided. Additionally, rise in temperature owing to vigorous stirring was also counteracted. High-Pressure Homogenisation by Microfluidizer®

For production of fine emulsions, a Microfluidizer® 110S high-pressure homogeniser (Microfluidics International Corp., MA-Newton, USA) was used, similar to those already described by Washington and Davis [1988], Lidgate et al. [1989] and Chaturvedi et al. [1992]. A preset pressure of up to 5 bar ( 75 psi) is amplified via a hydraulic assembly by a factor of 243 up to 1200 bar ( 18000 psi) hydrostatic pressure in the fluid circuit. The disruption of the samples is a result of the characteristic flow profiles of partitioned fluid streams that are reunited under high pressure and speed in the so-called ‘interaction chamber’. The disperse phase is disrupted by the high turbulence, shear and cavitation action occurring there [Schubert, 1997]. The dispersity obtained is directly dependent on flow velocity and thus homogenisation pressure applied. The homogeniser allows recycling of the samples for repeated dispersing action. To prevent excess heating of the sample during homogenisation, the liquid was driven through a heat exchange coil immersed in an ice bath, before re-entering the interaction chamber. Sample temperatures were thus maintained at about 30°C. No backpressure chamber (auxiliary processing module) was used with this model, as has been described elsewhere [Lidgate et al., 1989]. Nevertheless, dispersing results were reproducible and similar to those already reported. Fig. 16 shows a cross-section of the dispersion zone of the Microfluidizer 110S®. As recommended for processing of low viscosity samples [Microfluidics International Corp., 1996], the ‘F12Y’ type interaction chamber, which possesses a flow channel with a diameter of 75 µm at the narrowest site, was used. The chamber was made of an aluminium oxide type ceramic, which showed no abrasion throughout the experiments.

Fig. 16 Schematic cross-section through the dispersion zone of the Microfluidizer®, the so-called interaction chamber (left) and enlarged view of the liquid jet flow
through the dissipation zone in the centre area of the interaction chamber (right)

This apparatus has been especially designed for lab-scale homogenisation with a total volume of only 12 ml and a dead volume of only about 1 ml required for operation. Thus, small batches could easily be produced, as well as larger volumes by simply refilling crude emulsion in the reservoir and gathering the homogenised product at the outlet without recycling. For the experiments carried out here, it proved to be particularly useful not to use the continuous ‘recycling’ mode, as in that case only a proportion of about 6 ml per ‘cycle’ were processed through the ‘interaction chamber’ at once, and some unemulsified droplets and foam were still found floating on top of the liquid level, which never could be cycled through the dissipation zone, when the homogenised product was recycled to the product reservoir. By processing the whole batch once through the machine, a ‘true’ homogenisation cycle was carried out and no crude, unemulsified remainders were found after homogenisation.

Production of liposomes was achieved accordingly from the aqueous lecithin pre-dispersions. As Mayhew et al. [1984] reported, thus almost exclusively SUVs could be produced, which could be verified by Cryo-TEM investigations (see Chapter 3).

For production of SLNs from waxy materials (e.g. cetylpalmitate), the cooling coil was immersed in a heated, thermostatted water bath to allow operation above the melting point of the wax (approx. 54°C). After the final cycle, the product was collected in a beaker located directly in an ice bath and kept stirred for another 30 min to let the wax droplets solidify. The use of an additional in-line auxiliary processing module to achieve better dispersion results does not yield improved performance for low-viscosity O/W-emulsions [Broesel et al., 1998].

Fig. 17 Schematic drawing of the homogenisation process using the Microfluidizer 110S®

As illustrated in Fig. 17, the crude dispersions (A) were filled into the reservoir (B) and cycled through the dissipation zone (C), cooled or heated, respectively, by passing the heat exchange coil (D) and then collected at the outlet (E) to be refilled in the reservoir and recycled. After the final cycle, the pH of the aqueous phase was adjusted with 0.01 N sodium hydroxide solution to the desired value. Sterilisation by Autoclaving

Samples were autoclaved in a Getinge GE 666 EC-1 (Getinge AB, Getinge, Sweden) autoclave with a LAB100 control unit using a fixed ramp heating and cooling program. Sterilisation conditions were recorded by a plotter and it was thus possible to verify that pharmacopeial requirements (121°C, 2 bar) were maintained for the selected treatment time. Cooling of the samples to 80°C was achieved after some 15 minutes and subsequently accelerated in iced water while being shaken.

2.2.3 Measurement of Particle Size

No single particle-sizing method can be capable of describing the complete particle size distribution of such complex submicron systems like the emulsions under examination here. Several methods were, therefore, used here, which however measure different properties of the dispersed particles to deduce their ‘size’, making the results obtained difficult to compare [Groves, 1984]. For more detailed descriptions of the methods applied, the reader is referred to Allen [1990]. Photon-Correlation Spectroscopy (PCS)

PCS covers the size range of about 5 nm - 3 µm and has, therefore, extensively been used for particle size analysis of submicron emulsions [Komatsu et al., 1995]. It has also been proposed for inclusion in a new monograph ‘Globule Size Distribution in Intravenous Emulsions’ for the next issue of the USP in the Pharmacopeial Forum [Ph.Forum, 1994].
Fluctuations of laser beam scattering intensity I(t) from suspended particles in motion are detected and the auto-correlation function G() calculated for the sampling interval () [Allen, 1990].

Comparison of G() with its theoretical behaviour, g(), enables calculation of the particles’ diffusion coefficient (D):

According to Eq. 6 and Eq. 7, a faster decay of the auto-correlation function is therefore observed for smaller particles. The diameter of the dispersed particles (d) can then be computed (assuming spherical particles) using the Stokes-Einstein equation [Weiner, 1984]:

with k, the Boltzmann constant, absolute temperature T and the dynamic viscosity of the medium, . The interference of the auto-correlation functions caused by polydispersity can be deconvoluted by cumulants analysis:

a0 yields an intensity-weighted mean diameter, the so-called ‘z-average diameter’, and a1 gives the ‘Polydispersity index(PI) which is an expression of the width of the particle size distribution, ranging between 0 (uniform) and 1 [Allen, 1990 / Mueller, 1991].
The analyses were carried out on a Malvern Autosizer-LoC with a 5 mW Diode-Laser emitting a wavelength of 670 nm and detection under a fixed angle of 90° coupled to a 7032 Multi-8 Correlator (Malvern, Herrenberg, Germany). The correlograms were evaluated using PCS for Windows V1.27 original software by Malvern. Each sample was recorded 5 times using 10 sub-run measurements at 20°C. Auto mode and cumulant analysis were used by default. To avoid multiple scattering effects, a maximum of 50-100 Kcounts/s was adjusted by dilution of the samples with sterilised and 0.22 µm-filtered glycerol/water (2.25 wt%). Liposomal dispersions could be measured in the undiluted state. Laser Diffractometry (LD)

The ‘Mie parameter’ () relates the particle size (r) and the wavelength () of the scattered light from dispersed particles:

For < 0.1 (as for small particles), the Rayleigh approximation holds, where scattered light intensity (I) is found to be proportional to the sixth power of the particle diameter (d). For  > 0.5, the Mie scattering theory applies, and I ~ d2 [Allen, 1990]. Furthermore, the angle of scattered incident light is inversely proportional to particle size. The scattered light intensity measured at defined angles can therefore be related to the particle size. Fraunhofer theory is frequently employed in conventional Laser Diffractometry and simplifies Mie scattering behaviour by assuming spherical particles larger than the wavelength of the incident beam (d  4). It disregards the particles’ optical properties, most importantly refractive index. If the refractive index is known, more complex calculations according to the Mie theory encompass scattering and absorption effects of particles smaller than the wavelength used (d about /6) [Plantz, 1984]. Modern Laser Diffractometers can size particles from approx. 100 nm – 2000 µm in size using either Mie or Fraunhofer calculation. The former allows sizing of smaller particle classes, but requires knowledge of refractive index of the dispersed phase, which in the case of phospholipid-stabilised emulsions can only be approximated. The latter is more sensitive to larger particles [Lucks, 1993 / Kohlrausch and Steffens, 1997].
A Coulter LS 230 (Coulter Electronics, Krefeld, Germany) equipped with PIDS technique and running in Mie mode was used, as well as a Malvern Mastersizer Micro (Malvern, Herrenberg, Germany) running in Fraunhofer mode. 100 µl of the emulsion samples were diluted into approximately 120 ml glycerol/water (2.25% v/v). As with PCS, higher concentrations of liposomal dispersions had to be used to achieve similar scattering intensities (about 4 ml into 120 ml). The Coulter LS 230 uses a laser of 750 nm and a double Fourier lens setup for focussing the scattered light on the ring-shaped detector setup (Fig 18). The detection range (angles) for diffraction is claimed to be 40 nm - 2000 µm, thus theoretically covering the size range expected here. The detector segments are arranged in three main areas: a low-angle detector (L) with 62 detector elements, mid- (M) and high-angle (H) detector comprised of 32 elements each, allowing high-resolution of a wide particle size range.

Fig. 18 Assembly of the Coulter LS230 Laser Diffractometer
(adapted from Schoofs [1990])

The PIDS-technique (Polarisation Intensity Differential Scattering) is used in-line with the conventional LD sample cell. Horizontally or vertically polarised light of three different wavelengths (generated by a filter wheel) is scattered by the sample and recorded at six different angles (Fig. 19). PIDS is claimed to be especially sensitive for scattering caused by the smallest particles and intended for additional submicron size analysis [Bott and Hart, 1991]. Using the Mie theory for evaluating both LD and PIDS scattering data, particle sizes for liposomes and oil droplets should be resolvable from a single sample measurement.

Fig. 19 Schematic drawing of the PIDS-cell of the Coulter LS230
(adapted from Bott and Hart [1990]) Asymmetrical Flow Field-Flow Fractionation (AFFF)

FFF is a separation method similar to liquid chromatography, but with no stationary phase in the separation channel (Fig. 20). A force field perpendicular to the channel flow concentrates the sample components in the direction of the lower accumulation wall according to their size. Asymmetrical Flow FFF uses hydrodynamic flow as the separating force, with diffusion back into the channel acting as a counter force. When both are balanced, particles are retained at a defined height in the channel according to their size: larger particles accumulate in thin, compressed layers at the walls (Y in Fig. 20) and smaller particles are gathered in thick, diffuse layers towards the centre of the channel (X in Fig. 20). In the parabolic flow profile between the channel walls, the sample components closer to the accumulation wall are transported slower than those higher up in the channel, causing smaller components to elute first. The lower channel wall is made of a frit covered with an ultrafiltration membrane, which is impermeable for the sample. The channel flow divides into one part eluting at the outlet to the detector and one part exiting as cross flow. The ratio between both flows is controllable by adjusting the back pressure on the outlets. Thus, AFFF allows high resolution separation of the sample components according to hydrodynamic radius and, accordingly, particle sizing is possible [Giddings, 1993].

A light scattering detector was used to determine on-line radii of the eluted fractions. Light scattering intensity was measured at 18 angles simultaneously as a function of time. From the angular dependence of scattering intensity the mean square radius can be calculated from first principles, if the Rayleigh-Debye-Gans approximation holds [Wyatt, 1993]. This has been assumed in all calculations and is justified by the scattering properties of both, liposomes and emulsions [van Zanten et al., 1991]. A UV detector was also used, but gives a complicated mix of absorption and scattering effects, which makes it highly non-linear to concentration. Sizes can be calculated without any concentration information from light scattering alone, and knowing the size and the refractive index increment, also concentration can be calculated from the response of the light scattering detector [Wyatt Technology, 1998].

The experiments were carried out on an AFFF channel as described by Tank and Antonietti [1996], equipped with an ultrafiltration membrane (10k cutoff, Hoechst AG, Frankfurt, Germany) and coupled to a DAWN® laser light scattering detector (Wyatt Technology Corp., Santa Barbara, USA) and a Hewlett Packard 1100 UV detector (Hewlett Packard, USA). Samples were eluted in water containing 50 mM NaNO3.

Fig. 20 Principle of separation in a Flow FFF channel

2.2.4 Structural Analysis Centrifugation

High-speed centrifugation was performed on a Centrikon T-42K thermostatted centrifuge (Kontron Instruments, Neufahrn, Germany). Samples were pipetted into Beckman polyallomer tubes (No. 331374) and centrifuged using either an A-20 C type fixed-angle rotor at 19000 rpm (= approx. 33000 g) or a O-61 BC type swing-out-bucket rotor with a customised adapter to fit the tubes in the sample holders at 9000 rpm (= approx. 13000 g). Sample temperature was preset to 10°C and usually increased by about 4 - 5°C during centrifugation.

Ultracentrifugation was carried out on a Beckman model L centrifuge using either a 50 Ti fixed-angle rotor at 40000 rpm (= approx. 160000 g) or a SW 40 Ti swing-out rotor at 39000 rpm (= approx. 275000 g) and Beckman polyallomer tubes. Centrifugation temperature was also preset to 10°C, but in this case rose to about 20°C during centrifugation.

The polyallomer tubes were pricked with a syringe needle at the bottom of the tube to collect the subnatant (aqueous phase). Creamed oil phase and pelleted material could then be collected without further washing steps [Groves et al., 1985 / Férézou et al., 1994]. Freeze-Fracture Transmission Electron Microscopy (FF-TEM)

TEM can be used for the direct examination of particles in the size range 1 nm – 5 µm [Allen, 1990]. Thermal degradation or induced thermotropic phase transitions caused by the electron beam prevents, however, examination of samples containing lipids directly. The ‘slush technique’ was therefore used here. Samples were sandwiched between two gold platelets and cast into melting N2 (-210°C) prepared by evacuating liquid N2 with a vacuum-pump. Specimen vitrification with a cooling rate of 104-105 K/s was thereby achieved. Subsequently, the frozen samples were transferred into the Freeze-Fracture/Etching recipient (BAF 400 D, Balzers AG, Wiesbaden, Germany) under liquid N2 of -150°C using a special sampler holder. The samples were fractured at exactly -100°C under a vacuum of <10-6 mbar and subsequently shadowed with a 2 nm Pt layer (99.99% purity) at an angle of 45°. A final layer of pure carbon (20 nm) was added at 90°. Layer thickness was controlled by a swinging quartz assembly (QSG, Balzers AG, Wiesbaden, Germany). The replica were rinsed with chloroform:methanol (1:1 v/v), then with double-distilled water and transferred onto copper grids (400 mesh, Plano GmbH, Wetzlar, Germany). Finally, replica specimen were examined on a JEOL 100CX Transmission Electron Microscope using 80 kV acceleration voltage. Cryo-Transmission Electron Microscopy (Cryo-TEM)

This technique was developed and first introduced to allow visualisation and examination of liposomal dispersions and proved to be particularly useful for determination of size and lamellarity of these [Nattermann Phospholipid GmbH, 1995]. Clear advantages of this technique over other TEM techniques is, that native, untreated systems can be examined, thus lowering the risk of producing any artifacts, and furthermore, clear discernment between SUVs and small emulsified oil droplets was possible, which could not be achieved e.g. by FF-TEM [Westesen and Wehler, 1992].
Copper grids (200 mesh, Science Services, Munich, Germany) were coated with a porous film (Triafol BN, Merck KGaA, Darmstadt, Germany) and subsequently vapour-deposited with carbon. The Triafol film was removed by washing the grids with ethylacetate to obtain a pure porous film of carbon. The size of the holes, varying between 2 and 12 µm, depended on the preparation conditions. For specimen preparation, a drop of the sample (which was pre-diluted with glycerol-water or double-distilled water, alternatively by 1:11 v/v) was put on the untreated coated grid. Most of the liquid was subsequently removed by means of blotting paper, leaving a thin film stretched over the holes. The best hole coverage could be achieved using a hydrophobic film surface. The specimens were then vitrified by plunging them into liquid ethane cooled to 90 K in a temperature-controlled freezing unit using liquid N2 (Zeiss, Oberkochen, Germany). Excess ethane was removed from the specimen using blotting paper. Subsequently, the specimen were transferred into a cryo-transfer holder and inserted into a Zeiss CEM 902 transmission-electron-microscope equipped with a cryo-stage. Examinations were carried out at a constant temperature of 90 K using an acceleration voltage of 80 kV and 12 µA beam current. A condensor diaphragm of 100 µm and an objective entrance aperture of 17 mrad was employed. Zero-loss filtered images (DE = 0 eV) were taken under low-dose conditions using the minimal dose focusing device to avoid excess heating of the focussed areas. Polarised-Light Microscopy (PLM)

An Olympus IMT-2 microscope (Olympus Optical Co. Ltd., Tokyo, Japan) equipped with a photographic camera and crossed polarisers (/4 plates where indicated) was used. Lamellar phases exhibited typical anisotropic patterns and could thus be discerned from hexagonal or cubic phases which appear to be optically isotropic. Visual control of droplet sizes > 5 µm was also performed on this microscope. X-Ray Diffraction Analysis

Wide-angle X-ray Diffraction (WAXD) examination of samples was performed on an X’Pert X-ray diffractometer (Philips, Kassel, Germany). The samples were equilibrated for 30 min at the respective temperatures before data collection and were analysed under N2 purging. Thermostatting was achieved from 0°C - 90°C. 40 kV acceleration voltage and 40 mA current for generation of Cu-k radiation with  = 0.15419 nm (Ni-filtered) were used for recording diffraction patterns. 31P-Nuclear Magnetic Resonance Spectroscopy (31P-NMR)

With 31P-NMR it is possible to follow changes in motional properties of phospholipid molecules in membranes and lipid vesicles and gain information on the state of order in those systems, e.g. which phases (lamellar, micellar, hexagonal etc.) are exhibited [Cullis and DeKruijff, 1978 / Seelig, 1978]. This is possible since the differences in the spectra obtained originate from the gel or liquid-crystalline like behaviour of the lipids, which means more or less restricted ‘anisotropic’ motion due to the partially ordered structures taken within their aqueous dispersions. Commonly, ‘isotropic’ motion is understood to take a molecule over all directions in space with equal probability and is defined by a single characteristic rotational time, whereas with ‘anisotropic’ motion rotation about some molecular axes occurs at different rates, so that at least two characteristic rotational times must be used to characterise a molecule’s motion completely [Chan et al., 1981]. For non-bilayer arrangements (hexagonal phases, micelles etc.) additional motional averaging mechanisms lead to increased ‘isotropic’ movement, whereas typically, extended lamellar phases have been found to cause ‘anisotropic’ signals due to restricted motion in the bilayer plane [Cullis and De Kruijff, 1978]. One has to consider, however, that ‘motion’ as detectable by NMR encompasses motion within a molecule (e.g. trans-gauche rotations), motion of the molecule itself or even motion of larger molecular arrangements (e.g. tumbling of bilayer fragments or sheets). Due to overlayering of these effects, as a result, motional ‘averaging’ of the shift anisotropy might occur, leading to less information from the resulting spectra. However, 31P-NMR has been widely used in examination of biomembranes and model membranes (liposomes) and can be reviewed in more detail e.g. in Chapman [1975], Seelig [1978] and Chan et al. [1981].

A Varian Unity 300 spectrometer was used in the Fourier transform mode. Unless otherwise stated, the sample temperature was controlled to 250.5°C with a standard variable temperature control unit. In all cases of multi-temperature measurements the alterations in the NMR spectra were reversible upon recooling of the specimen. All chemical shift values are reported in parts per million (ppm) from pure LPC micelles (0 ppm), positive values referring to low-field shifts. All spectra were obtained in the presence of a gated broad-band decoupling (10 W input power during acquisition) and accumulated free induction decays were obtained from 1200 transients. In some cases, and in order to improve signal to noise ratio, higher transient numbers were used, which are given in the respective spectra. A spectral width of 25 kHz, a memory of 16000 data points, 1.3 s interpulse time and a 90° radio frequency pulse were used. Prior to Fourier transformation, an exponential multiplication was applied resulting in a 100 Hz line broadening. The residual chemical shift anisotropy, , was measured as 3 times the chemical shift difference between the high-field peak and the position of isotropically moving lipid molecules at 0 ppm. Fourier-Transform Infrared-Spectroscopy (FT-IR)

Fourier-Transform Infrared-Spectroscopy allows evaluation of infrared spectra with far better resolution and signal-to-noise ratio than with conventional spectrometers. It is thus possible to study multi-component samples and also follow temperature-induced peak-shifts owing to phase changes (e.g. melting or liquid crystalline phase transition of phospholipids). Especially carbonyl- and hydrocarbon-bands are susceptible for this approach [Fookson and Wallach, 1978 / Wallach et al., 1979 / Fringeli and Guenthard, 1981 / Casal and Mantsch, 1984]. This is achieved by recording and averaging high numbers of spectra, which is, when done conventionally, very time-consuming. Moreover, wavenumber accuracy is found to be in the order of better than 0.01 for FT-IR, allowing detection of even slight peak-shifts. Owing to its high resolution, even faint signals are detectable, and by computing the spectral data of multi-component systems, subtraction (of e.g. water from aqueous dispersions) or deconvolution of the respective underlying signals of the single compounds can be achieved [Davies, 1987].

A Nicolet Magna-IR 550 was used with Nicolet Omnic V2.0 software for data recording and manipulation. The measuring compartment was constantly purged with air free from CO2 and water by processing pressurised air through a KEN 6 TE type adsorption drier (Zander Aufbereitungstechnik GmbH, Essen, Germany). Different sample holders were employed which were mounted in temperature-controlled jackets and put on a motor-driven sample-shuttle to allow change of samples or background-samples. Temperature control between -10°C and 80°C was achieved by a Haake Fisons K 8752 type temperature bath connected to a second Pt-100 temperature sensor mounted directly into a small hole in the sample holder. To allow maintenance of equal sample thickness, conventional cuvettes from CaF2 with different spacers (Graseby Specac, obtained from L.O.T. Oriel GmbH, Langenberg, Germany) were initially used. Spacers thinner than 10 µm are, however, not readily available and difficult to handle. A customised ‘thin-film‘ cuvette (Fig. 21), also from CaF2, was therefore made. The latter was cut on one side to yield films of 5.6 µm thickness [Hienerwadel, 1993]. Samples were prepared by application of 0.6 µl of the liquid onto the sample compartment of the lower cuvette window (see Fig. 21) and subsequent tightening of the cuvette in a special sample holder. Any excess liquid was squeezed into the outer ring (overflow channel), allowing exactly reproducible sample thickness on each preparation.

Alternatively, attenuated total reflection (ATR/FT-IR) technique was employed at room-temperature, which could not, however, be exploited for quantitative data. Total reflection occurs when a beam hits an interface from the optically denser medium (n2 in Fig. 22) towards the optically ‘thinner’ medium (n1), provided the incident angle (2) exceeds a certain threshold angle. However, a part of the beam is able to penetrate into the ‘thinner’ medium for about a few wavelengths and then returns into the denser medium. By multiple reflection on the contact interface, the ‘thinner’ medium (i.e. the sample) thus is allowed to absorb respective bands from the incident beam. Since penetration into the sample is, therefore, inter alia dependent on material properties like refractive indices and wavelength, an amplification for peaks in the lower wavenumber region results in ATR spectra [Nicolet Instrument Corporation, 1993]. To be able to compare ATR/FT-IR data with conventional spectra, this effect was corrected by performance of an ‘ATR-correction’ on the ATR/FTIR spectra. Samples were analysed on a horizontal ATR contact sampling plate from ZnSe (Spectra-Tech Europe Ltd., Warrington, U.K.), which is also water-resistant, but allows measurements of absorbance down to about 650 cm-1 and possesses a refractive index of n = 2.4.

Fig. 22 Principle of Attenuated Total Reflection (ATR)/FT-IR (left) and of the horizontal ATR contact sampling technique (right) (adapted from Guenzler and Boeck [1993])

The analytical parameters and presets used for the FT-IR experiments were:




Beam Source

He-Ne Laser (1mW, 632.8 nm)







Spinning velocity





4 cm-1




None (or ATR-correction for ATR/FTIR-spectra) Differential Scanning Calorimetry (DSC)

DSC allows determination of thermotropic phase transitions in a quantitative manner and is especially useful for the investigation of the complex behaviour of polymorphic lipids [Small, 1986 / Gennis, 1989]. Samples were analysed on a Differential Scanning Calorimeter PL (Rheometric Scientific, Bensheim, Germany). Typically, 8 - 15 mg of the sample were weighed into an aluminium pan and cold-sealed. The reference pan remained empty and was sealed in the same fashion. Heating and cooling rates of 3°C/min (or as indicated) were applied, recording at least two complete heating/cooling cycles. Zeta Potential Measurement by Laser Doppler Anemometry (LDA)

For submicron parenteral emulsions, Zeta potential measurements via Laser Doppler Anemometry (LDA) are suitable. LDA allows fast determination of the electrophoretic mobility () using laser light scattering. The electrophoretic mobility is defined by the following equation, where E is the field strength [Volt/cm] and v the velocity of a particle along the field [µm/s]:

In the LDA setup (Fig. 23), a laser beam is first split and then interferes in the measurement zone located in the stationary layer of a capillary containing the sample. When particles following the electrical field move across this interference pattern, they cause a scattering of the laser light. This is frequency-shifted against the non-scattered parts of the beams (‘Doppler effect’). From the frequency-shift (or Doppler frequency, fd) measured, particle velocity through the capillary can be calculated according to Eq. 11, where is the laser wavelength and /2 is the angle of detection by the photomultiplier.

Fig. 23 Principle of Zeta Potential measurement by Laser Doppler Anemometry
as employed in the Malvern Zetasizer series (after Mueller [1996])

The electrophoretic mobility is calculated from Eq. 10 and then the Zeta potential from the Henry equation. Alternatively, the Helmholtz-Smoluchowski approximation is used by the Zetasizer 3000, as the particles are assumed to be large compared with the thin electrical double-layer [Hunter, 1981].

In this approximation, the Zeta potential () is derived from and the electric field strength (E) applied, as well as the viscosity () and the dielectric constant () of the dispersion medium at a given temperature.
Zeta potentials were determined using a Malvern Zetasizer 3000 (Malvern Instruments GmbH, Herrenberg, Germany) and are given as the mean of three measurements. Samples were diluted as described for PCS measurements (see Section, except that double-distilled and filtered (0.22 µm) water adjusted to the desired pH with 0.01 N HCl or NaOH solution was used for dilution. Measurements were carried out at 25°C  0.2°C unless stated otherwise. Calibration with a standard buffered sample (55 ± 5 mV) supplied by the manufacturer was performed before each set of measurements. Measurement of Conductivity

The conductivity of samples was determined using a customised assembly. A small vessel (approx. volume = 1 ml) contained two electrodes in fixed position connected to a Metex M-3650 Multimeter. When W/O type systems or oil as the sole phase were examined, no conductivity or infinite resistivity occurred.

2.2.5 Compositional Analysis Lyophilisation

Samples (5 ml of emulsions or lecithin dispersions containing approx. 1 - 4% lecithin which was equivalent to about 7.810-5 mol) were pipetted to steel freeze-drying cups ( 6.5 cm) and transferred into a Delta 1-24 KD Model freeze-drier (Christ Medizinischer Apparatebau, Osterode a. Harz, Germany). The following parameters were used:



Shelf temperature during freezing


Freezing time

2 h

Primary drying temperature


Primary drying time

30 h

Primary drying pressure (Pirani)

0.16 mbar

Secondary drying temperature

-10°C to 15°C at 3°C / h

Secondary drying time

10 h

Secondary drying pressure (Pirani)

0.14 to 0.16 mbar High-Performance Liquid Chromatography (HPLC)

Regarding the analysis of phospholipids by HPLC, numerous methods using various solvent mixtures have been reported, but owing to the complex nature of lecithins it is not straightforward, though. McCluer et al. [1986] give an overview of various methods and detection techniques applied. Most of those report utilisation of silica columns with isocratic elution, e.g. IUPAC [Beare-Rogers et al., 1992] proposed a method utilising a silica column, isocratic elution with n-hexane:2-propanol:acetate buffer (8:8:1 v/v/v) and detection at 206 nm. Taking into account that the substances intended to be determined in this case consisted of unknown quantities of rather complex mixtures of phospholipids and their more hydrophilic degradation products, it seemed appropriate to use a method that allowed determination of all components in a single run without having to sacrifice peak resolution. Furthermore, it was desired to limit sample preparation and manipulation to a minimum extent and to allow analysis of lecithin-triglyceride-glycerol mixtures as they occur in the emulsions. Therefore, utilisation of an eluent gradient seemed necessary, which in turn limited detection possibilities, since e.g. refractive index (RI) detection is bound to isocratic eluents [McCluer et al., 1986]. Sotirhos et al. [1986b] describe the separation of a variety of soybean lecithin components using an eluent gradient and UV detection at 210 nm. This method seemed suitable to resolve the complexity of the samples in question and was found to give similar retention times as they had been reported.

The analyses were performed on a Perkin Elmer system running Turbochrom V4.1 software (Bodenseewerk Perkin-Elmer GmbH, Ueberlingen, Germany). This comprised an ISS 200 autosampler and a Series 200 LC binary pump, with sample detection using a Perkin Elmer model 785 UV/VIS detector. Solvent mixtures were purged with He and kept in gas-tight Schott flasks with tightly mounted purging and solvent pipelines through the cap. This was to prevent evaporation of solvents and decomposition of gradient ternary mixtures during analysis, which had been found to be critical for reproducibility. Preliminary experiments had shown that direct admixing of solvents with a ternary pump did not work due to volume contraction of the solvent mixture leading to unstable baselines caused by bubbling. Calibration was carried out by injecting and evaluating weight/area plots of different solutions of PC (Lipoid EPC®), PE (Sigma), LPC (Sigma) and LPC (Lipoid ELPC®) in (2-Propanol : n-Hexane / 1:1 v/v). Relative standard deviations between three successive injections were found to lie in the range of about 1.5% for higher concentrated, well-responding compounds like PC or PE and about 3% for low-response signals from LPC. These values were comparable to the reports of Sotirhos et al. [1986b], except LPC which was reported to be detected with a relative error of about 10%. In preliminary experiments n-hexane : 2-propanol (1:1 v/v) was found to best dissolve all of the different substances of interest. Samples were therefore dissolved in 2-propanol : n-hexane (1:1 v/v) to give fixed volumes of injectable solution, whereby aqueous samples had been lyophilised in advance. In the case of more polar LPC, short and gentle heating helped to dissolve even larger amounts. Due to this fact, all samples were dissolved in the same fashion and thus could be injected using the same solvent mixture. To avoid sample losses, no filtering of sample solutions prior to injection was applied, since the use of a precolumn gave satisfactory results. Each analysis comprised three injections from separate flasks. During each experiment an additional sample of each respective standard solution was also analysed. Even after 500 injections no backpressure increase was observed during HPLC analysis.




30cm Waters µ-Porasil® 3,9mm ID (silica, 10µm size)
Precolumn: Waters Guard Pak® with Resolve® Silica


Eluent gradient

n-Hexane : 2-Propanol : Double dist. Water
(6 : 8 : 0.5) (v / v / v)
n-Hexane : 2-Propanol : Double dist. Water
(6 : 8 : 1.5) (v / v / v)


Over 20 min.: 90% + 10% changing to
+ 100% (linear)
For another 40 min.: 0%
+ 100%

Eluent flow

0.8 ml/minute

Injected Volume

75 µl


UV, 210 nm, 0.0010 AUFS

Fig. 24 shows a typical chromatogram for a commercial parenteral emulsion (Intralipid 30®). It can clearly be discerned between various lipid and phospholipid components which were not all determined or evaluated quantitatively in this studies.

Fig. 24 HPLC-Chromatogram of Intralipid 30®, showing main components and respective retention times High-Performance Thin-Layer Chromatography (HPTLC)

HPTLC compared to conventional TLC offers improved sample separation and requires only little equipment and time. Because other methods often have exhibited certain limitations and HPLTC offers a wide range of choices in stationary and mobile phases as well as in detection means, this technique is widely used in quantitative phospholipid analysis, e.g. by the forthcoming monograph of soya lecithin in the Ph.Eur. 1998. Two-dimensional HPTLC was employed to verify the lyso-phospholipid contents obtained from HPLC analysis. Sample solutions (2-Propanol : n-Hexane / 1:1 v/v) were pipetted on HPTLC plates using 10 µl DESAGA ‘minicaps’ microcapillaries (DESAGA, Heidelberg, Germany). On each plate three standard-solution dilutions (2-Propanol : n-Hexane / 1:1 v/v) were also applied and used for calibration. The trough was lined with filtration paper and allowed to saturate for 30 min. Elution was completed and the plate removed from the trough when the solvent front had travelled to about 1 cm away from the upper edge. The plate was dried with a fan and subsequently placed in the staining solution for 10 s. After drying away excess staining solution, the plate was developed in a conventional drying oven. After cooling to ambient temperature, the plate was scanned lane by lane using a Shimadzu CS-9301 PC Flying Spot Scanning Densitometer (Shimadzu Europe, Duisburg, Gemany). This kind of densitometer is especially suited for scanning of irregularly shaped spots and is also capable of evaluating data of non-homogeneously distributed spots [Marmer, 1985]. Spot absorption was compared with a fitted function of standard absorptions using the Shimadzu original software for quantitation. Content values are reported as the mean of three runs.



HPTLC plate

Merck silica 60 F254 HPTLC plates (10x10cm)


Chloroform : Methanol : NH3 conc. : Double Dist. Water 75 : 25 : 2 : 2 (v / v / v / v)

Sample Volume

3 x 10 µl

Staining agent

CuSO4-solution (10% w/v) in H3PO4 (6.8% v/v)

Developing conditions

45 minutes at 170°C


510 nm, reflectance mode, zig-zag scanning, 0.1 mm stepwidth

Fig. 25 shows a typical HPTLC-chromatogram after staining and scanned densitometric spectra which were integrated and quantitated subsequently.

Fig. 25 HPTLC-Chromatogram of Intralipid 20® (left, lanes 3 and 5) and densitometric
spectra thereof (right) showing lanes 3 () and 5 (-) from the chromatogram

Under the conditions mentioned above the following approximate Rf-values were observed:

Rf-values observed

Triglycerides (TG)


Free Fatty Acids (FFA)


Phosphatidylethanolamine (PE)


Phosphatidylcholine (PC)


Sphingomyelin (SPM)


Lysophosphatidylcholine (LPC)

0.10 pH-Measurement

A digital pH-meter type 640 (Knick, Berlin, Germany) with a glass electrode was used. Daily calibration with standard buffer solutions (pH 2.00 and pH 7.00 from Mettler Toledo, Steinbach, Germany) was employed. Measurements were carried out at room temperature on undiluted samples. See Chapter 3 for discussion of the measurement procedure.

Previous PageNext Page

Konvertiert vom Dissertationen Online Team im CCC der Universität Erlangen