Previous PageNext Page

Chapter 3 – Results and Discussion (I)

Production of Lecithin-Stabilised Emulsions

In this chapter, the physico-chemical characterisation of the egg lecithins and an examination of their emulsifying properties are given, followed by a step-by-step investigation of production parameters and their influence on emulsion size distribution, chemical composition and autoclaving stability.

3.1 Characterisation of Emulsifiers

For subsequent calculations, the molecular mass of the emulsifiers was estimated from the manufacturer’s declaration on phospholipid content and fatty acid composition.

Emulsifier

Molecular Mass (g/mol)

DPPC

734.1

Lipoid EPC®

770.5

Lipoid E75®

~762

Lipoid E80®

~762

Egg-PE

728.4

Egg-LPC

513.9

Sodium Oleate

304.4

 

Tab. 7 Molecular mass estimated for the emulsifiers used (values calculated according to data from Marsh [1990])

3.1.1 Characterisation of Emulsifying Properties

3.1.1.1 Measurement of Droplet Coalescence Times

The influence of emulsifier film properties on parenteral emulsion stability using a coalescence cell has already been described by Hansrani [1980]. She reported differences in coalescence times depending on emulsifier concentration and composition, but used the lecithin dispersed in the water phase and did not investigate the influence of elevated temperatures on coalescence behaviour. In this work the phospholipids were incorporated in the oil phase, allowing equilibration at the interface to be examined.
Various empirical evaluations of experimental data on film-thinning and coalescence of droplets at a planar surface have been proposed [e.g. Cockbain and McRoberts, 1953 / Gillespie and Rideal, 1954]. Typically, the resulting droplet rest times before their coalescence into the oil phase show a wide distribution, which can be evaluated according to the Cockbain-McRoberts equation (Eq. 13). This assumes a two-stage process, the first being the thinning of the liquid film to its critical thickness, and the second, faster one, the actual coalescence following first-order kinetics:

The first-order half life of droplet coalescence (T½) is calculated from the slope (‘coalescence rate constant’, kc) of a semi-logarithmic plot of the number of droplets not coalesced ()/() against time t. Accordingly, the film drainage time (td), where no coalescence has yet taken place, can also be obtained:

Fig. 26 Schematic of Cockbain-McRoberts plot for droplet rest-times

Hansrani’s [1980] coalescence data showed that the addition of minor components like LPC and SPM led to reduced coalescence times compared with pure PC. PE also showed a marked effect, although different PE concentrations in PC:PE mixtures did not produce large differences in coalescence time. Most pronounced effects were observed with addition of LPC to the PC:PE premix, followed by SPM. However, none of the compounds attributed to exhibit stabilising effects by charge repulsion (e.g. PS or PA [Rydhag, 1979 / Szoka and Papahadjopoulos, 1980]) increased droplet coalescence time. Hansrani used a 10-4 molar dispersion of PC:PE mixtures in water. Since, however, pure soybean oil gives similar values for T1/2 (see later), it is likely that insufficient emulsifier was available at the interface at the time of coalescence in Hansrani’s work. Furthermore, the dispersed lecithin vesicles she used would take a certain time to reorientate at the free oil interface (if at all). In the experiments reported here, the lecithin was, therefore, dissolved in the oil phase before layering onto the aqueous layer. It was expected that the lecithin would be more readily available at the oil/water interface compared with the aqueous vesicular dispersion, thus avoiding excess lecithin dispersed in the water phase without contributing to the film properties at the oil/water interface. Repeated measurements after various times were carried out in order to allow observation of interfacial ‘ageing’ phenomena.
With pure soybean oil, a clear effect of temperature on droplet coalescence rates is detectable (Fig. 27): increasing temperature decreases droplet stability, since film thinning is accelerated. Addition of lecithin to the soya oil produced a slow increase in droplet rest times. However, it took several hours to reach equilibrium rest times.

Fig. 27 (A) Droplet rest times of pure soya oil droplets in a Cockbain-McRoberts plot and (B) effect of prolonged equilibration times on T1/2 of 7.610-6 (O) and 3.310-6 molar () Lipoid E80® in soya oil at 50°C

Time to reach equilibrium was also dependent on lecithin concentration and took at least 48 h. Over a period of 72 h, a cloudy, viscous ‘layer’ gradually appeared at the interface, also dependent on the lecithin concentration. This ‘layer’ was assumed to consist of hydrated phospholipid (multilayer) structures. This observation supports the assumption that time-dependent life times are not just a result of slow diffusion of lecithin molecules through the oil phase. Remarkably, the T1/2 values determined for pure soya oil (Tab. 8) are of the same magnitude as of those obtained for an aqueous 10-4 molar PC dispersion reported by Hansrani [1980]. As can be seen in Fig. 28, increasing concentrations of Lipoid E80® led to increased coalescence stability when measured after the same equilibration time. With 2.610-5 mol Lipoid E80®, coalescence times at 70°C could be determined, whereas with 1.410-5 mol coalescence was too fast to be measured (Tab. 9).

Fig. 28 Cockbain-McRoberts plots for (A) 1.410-5 mol and (B) 2.610-5 mol of Lipoid E80® at different temperatures after 12 h equilibration

Lipoid E80® (1.410-5 mol)

T1/2 (s)

Lipoid E80® (2.610-5 mol)

T1/2 (s)

20°C

9.3

20°C

8.4

30°C

7.0

30°C

9.5

40°C

5.4

40°C

8.3

50°C

3.3

50°C

5.7

70°C

-

70°C

1.3

Tab. 9 First-order droplet coalescence half-lives (T1/2) for different Lipoid E80® concentrations at various temperatures after 12 h equilibration time

In order to obtain comparable droplet rest times of reasonable magnitude, phospholipids were employed in the range of some 10-6 mol/10 g of oil in subsequent experiments, and coalescence experiments were carried out at 50°C after 48 h equilibration (compare Fig. 27). It became evident, however, that the lecithin formed additional layers at the oil-water interface, which contributed to enhanced droplet rest times. According to reports on the emulsification properties of pure PC [Yeadon et al., 1958 / Rydhag and Wilton, 1981], and similar to the findings of Hansrani [1980], pure PC from egg yolk (Lipoid EPC®) gives more rapid coalescence than commercial egg lecithins containing increased content of minor compounds (Lipoid E75® and E80®) (Fig. 29). The latter contains approx. 80% PC and has a PC:PE ratio of 10:1, whereas Lipoid E75® possesses a PC:PE ratio of approx. 5:1 since about twice the amount of PE is contained. Yet, no difference in coalescence behaviour between both types is observed in Fig. 29 and Tab. 10.

Fig. 29 Cockbain-McRoberts plots for different commercial egg lecithins (7.710-6 mol each) at 50°C after 48 h equilibration time

Fig. 30 shows the influence of increasing LPC content measured for droplet coalescence times on pure PC films.

Fig. 30 Cockbain-McRoberts plots for different PC : lyso-PC (LPC)- mixtures (Lipoid EPC® : Sigma-LPC): (A) 7.710-6 mol PC, (B) 7.110-6 mol PC + 1.210-6 mol LPC, (C) 5.210-6 mol PC + 3.910-6 mol LPC at 50°C after 48 h equilibration time

A clear effect of LPC admixture is evident (Tab. 11), which is not simply a result of increased total lipid concentration. A marked increase in T1/2 is observed for 14.5 mol% LPC content and 42.9 mol% LPC in the mixtures, which is in agreement with Hansrani [1980], who did not find a levelling of LPC effect as in the case of PE content. Improved film properties were suggested, and possibly increasing stability of emulsions after partial hydrolysis of the emulsifier to LPC had taken place. Kumar et al. [1989] had also reported 31P-NMR data indicating a stabilising effect of LPC on SUVs from PC. Despite the problems with Hansrani’s method, it is clear that LPC improves droplet lifetime very effectively.

With both Lipoid E75® and E80® a much smaller effect of added LPC on coalescence time could be observed (Fig. 31). The ‘minor components’ already contained in Lipoid E75®/E80® evidently contribute to increased film stability. Increasing values for T1/2 were, however, observed on addition of LPC (Tab. 12), there being no difference between the Lipoid® types, as seen before. Again it is clear that LPC increases droplet lifetime, even for impure Lipoid® E75 and E80, although, for a more pronounced effect, relatively high content of LPC was required.

Fig. 31 Cockbain-McRoberts plots for () 7.710-6 mol Lipoid E75®, () 7.310-6 mol Lipoid E75® + 1.210-6 mol LPC and () 4.610-6 mol Lipoid E75® + 3.910-6 mol LPC (A) and for () 7.810-6 mol Lipoid E80®, () 7.110-6 mol Lipoid E80® + 1.210-6 mol LPC and () 5.210-6 mol Lipoid E80® + 3.910-6 mol LPC (B) at 50°C after 48 h of equilibration time

Lipoid E75®

T1/2 (s)

Lipoid E80®

T1/2 (s)

+ 0 mol% LPC

15.9

+ 0 mol% LPC

16.5

+ 14.1 mol% LPC

16.8

+ 14.5 mol% LPC

17.7

+ 45.9 mol% LPC

27.4

+ 42.9 mol% LPC

23.7

Tab. 12 First-order droplet coalescence half-lives (T1/2) for LPC admixtures on Lipoid E75®/E80® at 50°C

Muehlebach et al. [1987] determined 4.5 - 8.2% (presumably wt%) lyso-phospholipids by thin-layer chromatography in commercial Lipovenoes® and Intralipid® samples. Herman [1992] reported values corresponding to approx. 12 mol% LPC and about 4 mol% LPE for Intralipid 10%® as determined by HPLC with flame ionisation detection. For such LPC concentrations, however, the results in Tab. 12 predict only an insignificant effect on coalescence behaviour. It is concluded that LPC in commercial lecithins is unlikely to be responsible for reducing droplet coalescence, thus confirming Herman [1992], who did not observe significant changes in emulsion stability after addition of lyso-PC, rather than Hansrani’s prediction.
Sodium oleate is added as a co-emulsifying and pH-adjusting agent in certain commercial parenteral emulsions (see Section 1.2) and might also be formed during hydrolysis of lipids during production and storage. The influence of increasing amounts of sodium oleate (added to the aqueous phase) on the coalescence properties of PC-films is shown in Fig. 32.

Fig. 32 Cockbain-McRoberts plots for different PC (Lipoid EPC®) : sodium oleate mixtures: (A) 7.710-6 mol PC, (B) 7.810-6 mol PC + 0.710-6 mol sodium oleate, (C) 5.510-6 mol PC + 3.310-6 mol sodium oleate at 50°C after 48 hours

Again, coalescence is retarded by addition of sodium oleate (Tab. 13), but with less efficacy than observed for the addition of LPC. In general, it can be deduced that droplet coalescence is strongly influenced by the presence of ‘minor components’. Pure PC gives the least stable films, whereas sodium oleate and especially LPC both give enhanced coalescence stability. Rydhag [1979] reported increased lamellar phase swelling when sodium stearate was added to lecithins, which is also a possible explanation for the stabilising effect of oleate addition observed in Tab. 13. Hydrolysis of phospholipids during production and storage of the emulsion would intuitively cause increased coalescence stability owing to emulsifier film stabilisation. Pronounced effects could, however, only be observed when large amounts of additives were present. Since coalescence occurs faster at higher temperatures (Tab. 9), it seems that stabilisation owing to mere film rigidity does not occur under high temperature stress (during autoclaving), but might become more important under milder conditions (e.g. during pre-emulsification or storage). In all cases, droplet lifetimes increased with equilibration time until a plateau was reached, showing that stabilisation by lecithin emerges slowly. Thus lecithin should exhibit weakest stabilisation efficacy where fast surface coverage is needed, e.g. during homogenisation using high pressure. This is of vital interest when considering the results of subsequent structural studies.

3.1.1.2 Measurement of Film Compressibility

The apparent surface area (A0) covered by a single molecule [Å2/molecule] can be obtained from isothermal film compressibility measurements when plotting area (A) against film pressure () [mN/m].

Fig. 33 -A isotherm for an insoluble monolayer (after Hiemenz [1986])

For such data, Harkins [1952] discerned between gas-like molecules within a monolayer (G) which become more oriented towards each other upon increasing compression and are converted into the liquid expanded state (L1-G). This is a two-phase state where liquid domains coexist with ‘gaseous’ molecules. Upon further increase in pressure, the molecules are forced further together to form a liquid expanded film (L1). The intermediate state (I) is assumed to consist of crystalline and ‘liquid-like’ domains coexisting. Further increase in pressure causes the chains to become strongly oriented (crystalline), the molecules still being tilted from the surface normal (L2 or ‘liquid-condensed phase’). Finally, the solid state (S) is reached, where the tightest possible packing for the headgroups and chains is achieved. Excess pressure on such films leads to eventual breakdown of the film (Film Collapse Point, c) [Gaines, 1966 / Hiemenz, 1986]. Molecular space requirements and film compressibility (Cs) (degree of deformation upon pressure applied) of film-forming, water-insoluble amphiphiles like e.g. lecithin can be obtained by extrapolating the linear parts of the -A isotherm graph to A0 (area requirement at zero film pressure) to obtain molecular area requirement. Cs can then be calculated from the slope of the linear regions of the -A isotherm [Gaines, 1966]

Fig. 34 -A isotherm for palmitic acid at 20°C

Fig. 34 shows the -A isotherm for palmitic acid, which was examined in order to evaluate film balance performance. A film of the S-type is seen, where compressibility for the films is very low, before the collapse pressure is reached. The area requirement per molecule (A0=19.9 Å2) is in good agreement with data from the literature [Schueckler, 1992 / Lauda, 1987]. The -A isotherm for DPPC (Fig. 35) shows a two-phase region, where L1 and G phases coexist (L1-G), and either a second phase change or, more probably, incipient displacement occurring shortly before the collapse pressure was reached. Cevc and Marsh [1987] stressed that unsaturated phospholipids like DOPC remain in the L1 state at 20°C even at high surface pressures, and show higher A0 compared with saturated homologues. They also remarked that the formation of L2 phases by saturated phospholipids also depends on temperature. The molecular space requirement A0 determined for DPPC in Fig. 35 is similar, albeit slightly lower than e.g. reported by Lance et al. [1996].

Fig. 35 -A isotherm for Di-Palmitoyl-PC (DPPC) at 20°C

In contrast to DPPC, all Lipoid® lecithins appear to consist of only a single L1-phase throughout the measurement range, since no discrete phase change is observed upon film pressure increase (Fig. 36). This behaviour is often found for unsaturated phospholipids (see above). This again supports the assumption that the distinct two-phase region observed for DPPC in Fig. 35 is a result of the fatty acid substituent profile, since pure PC having a large proportion of unsaturated fatty acids (Lipoid EPC®) shows L1 behaviour and also increased A0 (Fig. 36). No marked differences in overall phase behaviour of the films between Lipoid EPC® and the two less pure lecithins are seen in Fig. 36, except that molecular area requirement, parallel with the content of impurities, increases in the order Lipoid EPC® < Lipoid E80® < Lipoid E75®. Similar molecular area values were e.g. cited by Groves et al. [1985], who assumed a molecular area for lecithin of 62 Å2 for interface coverage calculations. Broader, less sharp slopes than with DPPC are found for all lecithins, which is in agreement with the findings of Lance et al. [1996] for Lipoid E80®. These findings can be explained by the nature of the insoluble lecithin films, which contain mixtures of different lipids of both saturated and especially unsaturated fatty acid residues. They are therefore less easily compressed to an ordered arrangement, thus possessing higher compressibility values observed for the unsaturated residue lecithins.

Fig. 36 -A isotherms for Lipoid EPC® (), Lipoid E80® (-) and Lipoid E75® (N) at 20°C

Collapse pressure values of all three egg lecithins are markedly lower than that for DPPC (approx. 70 mN/m). Differences to the values quoted by Lance et al. [1996] can be explained by the higher film compression rate used in their experiments, since collapse pressure is known to be very sensitive to the compression rate applied [Hiemenz, 1986]. However, in general the lower collapse pressure arose from less densely-packed films with less molecular interactions compared with DPPC. This is a result of differences in fatty acid residues (homogeneity and degree of saturation) rather than of different phospholipid components, since all egg lecithins behaved similarly despite their different composition. As film collapse is subject to various influences, information from it has to be regarded with due care [Gaines, 1966] and is only limited. The slight differences in collapse pressure values observed (in the order Lipoid EPC® < Lipoid E80® < Lipoid E75®) indicate, however, ideal miscibility of components, which excludes separation or ‘clustering’ into separate PE or PC domains [Gaines, 1966]. For the case of the oil-water interface, increasing content of impurities should result in slightly less rigid emulsifier films. The more rigid films from pure PC were, however, less stable to droplet coalescence (see Tab. 10).
Insoluble monolayers of PE are known to be more easily tightly packed than for PC [Chapman, 1975], which is related to its higher chain melting transition temperature. Also its head group area is smaller than that of PC. Increasing content of PE in the commercial lecithins along with other minor impurities appears to cause the observed increase in apparent molecular area requirement and compressibility (Tab. 14). Molecular interaction between different phospholipid species are apparently decreased in the monolayer films. Cs values of commercial egg lecithin films are expectedly higher than for the more rigid DPPC layers and for the pure egg PC (Lipoid EPC®). The value for Lipoid E80® is a little lower than reported by Lance et al. [1996]. But as their value for DPPC also ranged higher than the one given here, substantial agreement is assumed. Since, however, the Cs values of all Lipoid® lecithins were more similar to each other than e.g. compared with DPPC, the film properties alone do not explain, why Lipoid EPC® possessed remarkably worse emulsification properties than Lipoid E80® and E75®.
Electrostatic charge increases molecular area requirement owing to electrostatic repulsion [Hiemenz, 1986] and gives a possible explanation for the behaviour of Lipoid E75®/E80®, where more charged species are present and molecular area requirement is larger than with PC (Tab. 14). A similar explanation might apply for the case of higher subphase pH values (pH is adjusted to alkaline range before autoclaving emulsions), resulting in increased molecular area. However, this could not be investigated with the method used here, since Langmuir trough measurements require that the monolayer be completely insoluble in the subphase. Owing to hydrolysis and ionisation, fatty acids derived from the lecithin would dissolve at higher subphase pH values [Schueckler, 1992] and disappear from the monolayer film.
The molecular weight values used here for the molecular area calculations are taken from the declaration of composition and fatty acid profile by the manufacturer. Lance et al. [1996] estimated considerably lower values of molecular weight for Lipoid E80®, which accounts for the higher molecular area reported by these authors of 69 Å2 for Lipoid E80®. This deviation also possibly resulted from the different method used (Wilhelmy plate) by these authors. Molecular area requirement for lecithin is, therefore, not straightforward to determine. As discussed before, however, the complicated mixtures of Lipoid E75®/E80® behave like a single species, which allows assumption of an average single molecular area.

3.1.2 Characterisation of Emulsifier Composition

3.1.2.1 Emulsifier Composition according to HPLC

Gradient HPLC analysis of the emulsifiers using UV-detection was successfully employed, showing that Lipoid EPC® contains slight, but detectable amounts of an impurity, which is assumed to be sphingomyelin (SPM). Lipoid E75® shows considerably more phosphatidylethanolamine (PE) and slightly less phosphatidylcholine (PC) than Lipoid E80®, and both contained similar trace amounts of SPM and faint traces of lyso-phospholipids. Fig. 37 shows the chromatograms of Lipoid EPC®, Lipoid ELPC®, Lipoid E75® and Lipoid E80®. Retention times are similar to those of Sotirhos et al. [1986b], who also described peaks of PI, PA and PS appearing in the region between PE and PC. The minor peaks observable at retention times between 16 and 25 min were not examined more closely here, as these components are not listed in the analysis reports of the manufacturer. The peak-splitting or shouldering observed at high concentrations is attributed to the different fatty acid substitution profile of the phospholipids, as previously reported by Herman [1992]. In this case co-integration of peak and shoulder was performed. Since UV-detection of phospholipids is a response of carbonyl and unsaturated sites [McCluer et al., 1986 / Herman, 1992], lyso-phospholipids are expected to yield a weaker absorbance. This was confirmed by a markedly reduced slope of the LPC-calibration curve compared with PE (largest slope) and PC. No differences could be found between Sigma and Lipoid LPC material, which is as an indication of similar fatty acid substitution (unsaturated/saturated fatty acid ratio) for both LPCs. As stated by McCluer and co-workers [1986] and Herman [1992], values for lyso-phospholipids are especially susceptible to the species of fatty acid cleaved from the phospholipid, since loss of an unsaturated fatty acid leads to a larger decrease in absorbance compared with loss of a saturated fatty acid. The LPC peaks are also split in 3-4 peaks (Fig. 37b), even though there is only a single fatty acid residue. It was hoped that by using egg-yolk compounds as standards, the influence of content variability in fatty acids on detection could be minimised.
As could be seen in Tab. 15, the HPLC assay for PC and PE are similar to the HPTLC results given by the supplier of the lecithins. In the case of LPC, however, unrealistically high values are found, although Sigma LPC and Lipoid® ELPC both gave reproducible results on repeated injection. The LPC contents are much higher than the correspondent decrease in PC concentration on hydrolysis (see later). Herman [1992] stated that hydrolysis of fatty acids would lower the UV molar absorbance coefficient of LPC in an unpredictable manner, making it impossible to be quantified by UV detection.

Fig. 37 HPLC chromatograms of (A) Lipoid EPC® (with enlarged section), (B) Lipoid ELPC® (enlarged), (C) Lipoid E75® (with enlarged section) and (D) Lipoid E80® (with enlarged section)

Sample

PC (mol%)

PE (mol%)

LPC (mol%)

Others

Lipoid E75®

76.0

19.5

19.5

SPM detectable

Lipoid E80®

82.3

10.9

14.5

SPM detectable

Tab. 15 Results for HPLC determination of emulsifier composition

Since the UV absorbance of LPC is low, measurement error was high owing to variations in the remaining fatty acid residue. It could, however, not be clarified why almost tenfold LPC contents are found compared with the supplier’s declaration. LPC concentrations were, therefore, controlled by quantitative HPTLC measurements, as reported in the following section.

3.1.2.2 Emulsifier Composition according to HPTLC

Although only intended for lyso-PC, absorption values for other phospholipids could also be compared between different samples, allowing an estimation of compositional differences. As can be seen from Figs. 38 and 39, the HPTLC analysis demonstrated that Lipoid EPC® contains PC and minute amounts of SPM, but no detectable quantities of lyso-PC. However, for Lipoid E75® and E80® some LPC was clearly detectable. Lipoid E75® also contains more impurities in the form of PE and SPM than does Lipoid E80® (Fig. 39). Quantitation of each chromatogram was carried out by integration of the sample peak areas and using Sigma-LPC standard dilution stains.

Fig. 38 HPTLC chromatogram of Lipoid E80® (lane 1), Lipoid E75® (lanes 3 and 5) and Lipoid EPC® (lanes 7 and 8)

Fig. 39 HPTLC densitometric spectra of Lipoid EPC® (-), Lipoid E75® () and Lipoid E80® (N)

Tab. 16 summarises the results for LPC content of the lecithins, clearly indicating lower values for all samples than determined by HPLC analysis.

Sample

LPC (mol%)

Others

Lipoid EPC®

not detectable

SPM detectable

Lipoid E75®

3.1

SPM detectable

Lipoid E80®

2.2

SPM detectable

Tab. 16 Results for HPTLC determination of LPC content

HPTLC analysis thus shows that the LPC content is in the range of the supplier’s declaration, indicating that HPLC analysis yielded too high values for LPC and should not be used for LPC quantitation.

3.1.3 Physical Characterisation of Emulsifier

3.1.3.1 Thermoanalytical Characterisation

Comparison of lipid chain melting behaviour may be of importance for emulsification, where phospholipids are assumed to orientate more easily at the interface when in the molten state [Diederichs, 1993]. As chain length and degree of saturation of fatty acid residues have a strong influence on packing behaviour (see Chapter 1), calorimetric properties of phospholipids also vary with their composition. Marsh [1990] gives main transition temperatures for fully hydrated, saturated PCs, which increase with chain length from -1.1°C (di-lauroyl-PC) to 64.5°C (di-arachidoyl-PC). Similarly, the transition enthalpies also increase with chain length. For PE, higher transition temperatures (di-lauroyl-PE: 30.2°C / di-arachidoyl-PE: 82.0°C) and enthalpies are cited. For di-substituted PCs, Lc-L phase transitions occur at about 21.2°C for palmitic residues, 28.0°C for stearic residues and 20.5°C for arachidic residues. L-P pre-transitions lie at 34.2°C (di-palmitoyl-PC), 50.7°C (di-stearoyl-PC) and 63.7°C (di-arachidoyl-PC). L phases are formed by P-L phase transitions (‘chain melting’) at 41.4°C (di-palmitoyl-PC), 55.3°C (di-stearoyl-PC) and 66.4°C (di-arachidoyl-PC). However, for unsaturated or mixed saturated/unsaturated phospholipids less exact data is available. Typically, lower transition temperatures are observed for PC with unsaturated chains, e.g. -21°C (di-oleoyl-PC), -16.2°C (1-stearoyl-2-linoleoyl-PC) or -13°C (1-stearoyl-2--linoleoyl-PC). Diederichs [1993] reported transitions for dry soya-PC with mainly palmitic and stearic fatty acids at approx. 100°C, which is in agreement with e.g. Cevc and Marsh [1987] who quote 100°C and Small [1986], who quotes 110°C for the main transition of anhydrous DPPC. Diederichs also found a transition temperature for non-hydrated egg lecithin at around 25°C. Complex egg lecithin mixtures like Lipoid E80® should, therefore, show similar behaviour and exhibit broader transition peaks than more refined Lipoid EPC®.

Fig. 40 DSC thermogram of desiccated Lipoid EPC® (at 3K/min)

Fig. 41 DSC thermogram of desiccated Lipoid E80® (at 3K/min)

After drying under vacuum 12 h prior to examination, the unsaturated egg lecithins show melting slightly above room temperature (Figs. 40 and 41) as already reported by Diederichs [1993]. Nevertheless, Lipoid EPC® yields a narrower peak than Lipoid E80®, which shows a very broad transition stretching below 0°C. As the fatty acid profiles of both are comparable, this difference must be a result of the higher content of minor components present in the Lipoid E80®. In contrast to the reports of Diederichs [1993], who observed transition peaks at about 29.4°C and transition enthalpies of 24.5 mJ/mg, the latter were found to be 32.34 mJ/mg for Lipoid E80® and only 20.84 mJ/mg for Lipoid EPC®. However, according to the literature [Marsh, 1990], enthalpies of about 40 - 58.4 mJ/mg would be expected for PC containing unsaturated fatty acid chains (on the basis of the estimation of 770 mol/g). The deviations for the results reported here were probably owing to residual or re-adsorbed water content of the lecithins owing to their marked hygroscopicity [Small, 1986], as absorbed water may well account for the measurable lowering of the transition enthalpy. 7.2% w/w microfluidized liposomal dispersions in 2.25 wt% glycerol/water from Lipoid E80® show thermograms dominated completely by the sharp water freezing and melting peaks at -12.6°C and 2.6°C, respectively (not shown). No peaks arising from different lecithin concentrations could be observed. To shift the water peaks to lower temperatures and expose hereby the lecithin transitions, increasing amounts of glycerol were used in the dispersion medium. Thus, 30 wt% Lipoid E80® dispersed into 50 wt% glycerol/water (Fig. 42) shows a clear shift of the water freezing and melting peaks down to about -67°C and -31°C, respectively. This reveals broad solidification and melting peaks at about -11°C, similar in their width and shape to those observed for the desiccated lecithins. Similar behaviour has been reported for egg-yolk lecithin of unknown composition by Chapman [1975], who reported melting peaks at -15°C for the heating and -7°C for the cooling curve.

Fig. 42 DSC thermogram of 30 wt% Lipoid E80® in glycerol/water (50 wt%) at 3K/min

Diederichs [1993] reported broad transitions ranging from -35°C to +18°C with a peak maximum at about -10°C with 15% water content, and stressed that no peaks could be found at higher water contents. It is evident from Fig. 42 that the homogenisation temperature for emulsions stabilised with Lipoid E80® need not be above room temperature to yield an L phase of the fully hydrated lecithin. To control the influence of factors like emulsifier hydration, diffusion and hydrolysis, which are strongly dependent on the temperature, model emulsions should, however, be predispersed at elevated temperatures to allow fast hydration of the emulsifier.

3.1.3.2 Characterisation by FT-IR

Using FT-IR, subtle shifts of the hydrocarbon- or carbonyl absorption bands of phospholipids can be detected, especially the CH2-stretching and -scissoring modes reflecting the state of order in the acyl chains in the fatty acid residues [Fookson and Wallach, 1978 / Fringeli and Guenthard, 1981 / Casal and Mantsch, 1984]. Shifts towards higher frequencies, e.g. at 2850 cm-1 or 2920 cm-1, indicate an increasing proportion of gauche to trans conformers, as occurs for the chain melting at the main transition of the lipid chains [Casal and Mantsch, 1984]. It is therefore possible to assess phase transitions using FT-IR by recording temperature-dependent spectra. A well-characterised example is DPPC [Fookson and Wallach, 1978], where hydrogen-bonding shifts the phosphoryl band at 1254 cm-1 to 1248 cm-1. For DPPE, the same band appears at 1222 cm-1 owing to strong intermolecular hydrogen-bonding, which is also reflected in the spectra of mixed films of DPPC/DPPE, where the phosphoryl band lies in between those of the pure species and somewhat closer to the value of DPPC. The authors concluded that intermolecular hydrogen-bonding of the PE molecules is reduced and intermixing of both phospholipid classes becomes possible, especially at more alkaline pH where NH-O hydrogen-bonding becomes increasingly disrupted. Casal and Mantsch [1984] reported the main transition (chain melting) at 41.5°C for hydrated DPPC as determined from the CH2-stretching modes. The pre-transition at 35°C was determined using the CH2-scissoring modes. The carbonyl-stretching modes were also reported by these authors to be sensitive to the main transition, when head groups change conformation owing to chain movement. In the case of unsaturated egg-yolk PE, a L-HII (lamellar-inverted hexagonal) transition was detectable, which occurred at lower temperatures when more unsaturated fatty acids residues were present. Fig. 43a shows that non-desiccated Lipoid EPC® contains water (arrows indicate OH-modes caused by water) which could not be detected by DSC measurements. To dry films prepared from chloroform solutions, residual solvent and water were therefore removed by evacuating the sample for 12 h, which now showed the absence of water OH-stretching modes in Fig. 43b.

Fig. 43a ATR/FT-IR spectrum of non-dried Lipoid EPC® at ambient temperature

Fig. 43b FT-IR spectrum of desiccated DPPC from chloroform solution at 21°C

Tab. 17 gives an overview of the typical values of the bands most important for analysis of lipids and phospholipids examined here:

IR-Band

Wavenumber (cm-1)

C=O (sn-1 / sn-2) – stretching (in esters)

1742 / 1725

C-O – stretching

1170 + 1070

C=O – stretching (in acids)

1700-1725

CH2 – stretching, antisymmetric

2920

CH2 – stretching, symmetric

2850

CH2 – deformation (scissoring)

1465

CH2 – deformation (wagging)

1305

CH2 – deformation (twisting)

1180-1345

CH2 – deformation (rocking)

720

Terminal CH3 – stretching, antisymmetric

2956

Terminal CH3 – stretching, symmetric

2870

=C-H – stretching, antisymmetric

3010

CH3 – stretching in N(CH3)3, antisymmetric

3040

N-(CH3)3 – stretching, antisymmetric

970

C-N – stretching, antisymmetric

945

PO2 – stretching, symmetric

1085-1100

PO2 – stretching, antisymmetric

1220-1250

P-O – stretching, antisymmetric

815-825

-OH – stretching

3200-3600

Tab. 17 IR-bands of importance for phospholipid analysis
(after Fringeli and Guenthard [1981], Weers and Scheuing [1991] and Guenzler and Boeck [1993])

Fig. 44 shows the temperature dependence of the symmetric CH2-stretching mode of DPPC, which was dispersed in water and heated to 60°C, vortexed and frozen to -20°C. This freeze-thaw cycling followed by vortexing was repeated four times to achieve fully hydrated material. A clear shift of about 2 cm-1 is observed, indicating L-L chain melting at 39°C, which is the shift reported by Wallach et al. [1979] and Cortijo et al. [1982]. The deviation from the literature value of 41.5°C [Casal and Mantsch, 1984] is assumed to be a result of the slower heating rate of 0.2 K/min used here. The desiccated DPPC shows no shift in the range up to 80°C, as no transition is expected to occur below 110°C [Small, 1986]. This effect demonstrates the influence of water on the thermal behaviour of the phospholipids. Hydration of the head groups and subsequent head group packing rearrangement lead to a decrease in the transition temperature shift and its enthalpy [Cevc and Marsh, 1987]. The symmetric stretching mode has the advantage over the asymmetric band at about 2930 cm-1, that it is less easily overlapped with other modes [Casal and Mantsch, 1984]. For DPPC, the main transition could be observed from the concomitant shift of the asymmetrical CH2-stretching mode and the CH2-scissoring mode.

Fig. 44 Peak location for CH2-symmetrical stretching modes of desiccated (O) and fully hydrated DPPC ()

The commercial lecithins were dissolved in chloroform and dried as a thin film on a CaF2-window under vacuum for about 12 h. Figs. 45 and 46 show clear shifts of the CH2-stretching and -scissoring modes. The transition is, however, less sharp than that of hydrated DPPC in Fig. 44.

Fig. 45 FT-IR spectra of dry Lipoid E75® at 9-51°C with 3°C increment from bottom to top: region of the CH2-stretching modes

Fig. 46 FT-IR spectra of dry Lipoid E75® at 9-51°C with 3°C increment from bottom to top: region of the CH2-scissoring bands

Figs. 47 and 48 summarise the temperature-frequency dependence of the CH2-stretching and -scissoring modes for the different desiccated Lipoid® lecithin types. The wide transitions occur at approx. 43°C for Lipoid EPC®, 32°C for Lipoid E75® and 29°C for Lipoid E80® in desiccated samples. The desiccated lecithins possess, therefore, lower transition temperatures than dry DPPC (Fig. 44) owing to their higher content of unsaturated PC species. This is further lowered slightly by the higher content of impurities in Lipoid E75® and Lipoid E80®.

Fig. 47 Peak location for CH2-symmetrical stretching modes of desiccated Lipoid E75® (O), Lipoid E80® () and Lipoid EPC® ()

Fig. 48 Peak location for CH2-scissoring modes of desiccated Lipoid E75® (O), Lipoid E80® () and Lipoid EPC® ()

For Lipoid E75® and E80®, these chain melting transitions lie at similar values, although theoretically the ‘purer’ Lipoid E80® is expected to have a higher transition temperature than Lipoid E75®. Indeed, with Lipoid E80® the DSC peak maximum was 5°C lower. The residual water content in the DSC samples could not, however, be controlled as closely as with the FT-IR method. The main transition for Lipoid EPC® occurs 15°C higher than that determined from its DSC peak maximum (Fig. 40). This may again be a question of higher residual water content in the DSC samples, or simply reflect the difference between the two techniques.

Detection of the main transition for hydrated Lipoid E80® is shown in Figs. 49 and 50.

Fig. 49 Peak locations for CH2-symmetrical stretching modes of desiccated (O) and fully hydrated Lipoid E80® ()

Fig. 50 Peak locations for CH2-scissoring modes of desiccated (O) and fully hydrated Lipoid E80® ()

In contrast to the DSC results (Section 3.1.3.1), it is thus possible to detect the gel-liquid crystalline phase transition for the fully hydrated Lipoid E80®. For the symmetrical CH2-stretching mode an offset in the sigmoidal curve is apparent (Fig. 49), which can be assigned to the so-called ‘ice-melting-induced shift’, which usually is more easily detectable for modes which are strongly affected by hydrogen-bonding such as head group modes [Casal and Mantsch, 1984]. The detection of the gel-liquid crystalline transition is, however, not affected by this phenomenon and occurs at around 4.5°C, also about 15°C higher than the peak maximum determined from DSC data of the glycerol/water dispersion (Fig. 42). The DSC experiments gave, however, only a very broad transition ranging from -30°C to 10°C and thus do not enable assignation of distinct molecular mechanisms. From the FT-IR data, however, an unequivocal correlation between acyl chain mobility and temperature is achieved, showing that upon hydration of Lipoid E80® a strong shift of the main transition occurs. All of the lecithin is present in the L-phase above approx. 10°C. The differences found for transitions determined by DSC and FT-IR could partly be a result of the slightly slower heating rate used for FT-IR (about 0.2 K/min.), which might have caused the DSC peaks to appear at a few °C lower than determined from FT-IR spectra. One of the most striking advantages of FT-IR is that even hydrated egg lecithin samples can be examined, whereas DSC thermograms quickly become overridden by the strong influence of peaks arising from water [Diederichs, 1993]. By computational subtraction of the typical water bands, the chain and headgroup bands of even minor compounds are detectable with FT-IR.

3.1.3.3 Characterisation by X-Ray Diffraction

From x-ray diffraction spectra it is possible inter alia to discern between anisotropic and isotropic phases. Following the Bragg equation, one can calculate molecular spacings from the typical scattering angles observed [Cevc and Marsh, 1987]. In Fig. 51, the temperature-dependent x-ray diffraction patterns for desiccated Lipoid EPC® are depicted, showing reflections (especially a sharp peak at 20.3° = 4.85 Å repeat distance) indicating crystalline chain-packing at room temperature [Small, 1986 / Larsson, 1986], which still is the case at 22°C. At 52°C, it exists in the molten state with no three-dimensional repeat-distances occurring.

Fig. 51 X-Ray diffraction patterns for desiccated Lipoid EPC® at different temperatures

With Lipoid E80® (Fig. 52) and Lipoid E75® similar behaviour is found, confirming that complete melting of the lecithin crystals was achieved above 50°C. At 20°C, the diminished intensity of the reflection at 20.4° indicates incipient decrease in cristallinity.

Fig. 52 X-Ray diffraction patterns for desiccated Lipoid E80® at different temperatures

These results confirm the data from DSC and FT-IR investigations.

3.2 Preparation of Emulsions, Solid Lipid Nanoparticles and Liposomes

The influence of emulsifier composition, its total content and manufacturing conditions on product properties were examined from measurement of particle size distribution, Zeta potential and emulsifier composition during production. Commercial formulations are reported for comparison.

3.2.1 Limitations on Determination of Particle Size

The most frequently employed methods for submicron emulsion particle analysis nowadays are PCS and LD [Groves, 1984 / Komatsu et al., 1995 / Schuhmann, 1995]. Other authors reported higher sensitivity for the largest droplet fractions using Coulter Counter sizing [Washington and Sizer, 1992 / Schuhmann et al., 1993], which, however possesses only a limited measurement range (> 700 nm), and prerequisite sample dilution in saline solutions promotes emulsion instability and coalescence [Schuhmann and Mueller, 1998]. Westesen and Wehler [1993] stated that the ‘intensity distributions’ obtained from PCS measurements were not useful for calculation of number or volume distributions. Preliminary experiments in this work showed, indeed, that monomodal cumulant analysis always yielded log-normal distributions even for broad, polydisperse distributions as expected for the emulsions investigated. Fig. 53 shows that samples were fitted to a perfect log-normal distribution by cumulant analysis, which does not necessarily represent their true distribution shape. Liposomes present may, for example, form a second distribution which is not resolved. Only z-average values and polydispersity index (PI) are, therefore, reported here. Dilution of the emulsion samples did not alter the size distributions obtained as could be verified beforehand. PCS is limited to detection of particles undergoing Brownian motion [Mueller and Schuhmann, 1996], and particles larger than approx. 3 µm are therefore not detectable. As shown in Fig. 54, a visibly broken emulsion could not be correctly analysed with PCS. It was, nevertheless, possible to detect a pronounced increase in z-average and PI, which allowed coalescence to be qualitatively followed.

Fig. 53 PCS intensity distributions of a 20% soya oil emulsion (A), cumulative percentage frequency curve in log-scale (B) and cumulative percentage frequency curve in log-probability scale (C)

Laser Diffractometry (LD) determines particle distribution from the actual intensities on the detector elements. Thus multimodal, polydisperse samples can also be resolved with high-resolution. To examine the upper region of emulsion size distributions which are of special interest regarding their parenteral use, better sensitivity towards larger particles could be achieved by evaluating the diffraction data using the Fraunhofer approximation and presenting data in the form of volume distributions [Kohlrausch and Steffens, 1997]. However, more precise detection of the whole emulsion size range, especially at the lower size range, is achieved using the Mie approximation. As it is, however, difficult to determine the necessary values for the refractive index of the dispersed phase, Mie calculations were carried out assuming different refractive indices ranging from 1.33 (for water) to 1.59 (for polystyrene standard latex dispersions).

Fig. 54 PCS intensity distribution (A) and an LD (Fraunhofer) distribution (B) of a broken 20% soya oil emulsion

3.2.2 Influence of Emulsifier Concentration

The Microfluidizer® 110S is especially designed for lab-scale homogenisation with a volume of 12 ml and a dead volume of only 1 ml. Thus, small batches can easily be produced, as well as larger volumes by refilling crude product in the reservoir and gathering the homogenised product at the outlet. It proved useful not to use the continuous ‘recycling’ mode, as in that case only about 6 ml per cycle were processed through the interaction chamber, and some unemulsified droplets (in the case of emulsions) and foam were found floating on top of the liquid. Accordingly, this material could never be cycled through the dissipation zone. By processing the whole batch through the machine at once, no crude, unhomogenised material was evident. The collected sample was then refilled into the reservoir and recycled through the homogeniser.
Liposomes were prepared by microfluidization of the aqueous lecithin pre-dispersions. Mayhew et al. [1984] claimed that almost pure SUVs could thus be produced. PCS measurements of 1.2 wt% up to 7.2 wt% liposome dispersions were performed in their undiluted state (Fig. 55) and showed that particle size constantly decreases during homogenisation, yielding almost exclusively small liposomes. Particle size falls drastically after the first homogenisation cycle and drops more slowly upon recycling, until final z-average diameters are in the range of approx. 50-120 nm after 11 cycles. Similarly, the width of size distribution detected by PCS (PI) dropped over 3 cycles, then remained constant at about 0.64, which indicates broadly distributed dispersions.

Fig. 55 Dependence of z-average diameter and polydispersity index of 7.2 % w/w Lipoid E80® (liposome) dispersions [homogenised at 1000 bar; n=5]

Fig. 56 shows the volume distributions for the example of a 3.6 wt% Lipoid E80® dispersion obtained from LD using PIDS and Mie calculation. Different assumptions of refractive index for the Mie calculation have an effect on the distribution, although not very pronounced. SUVs are, therefore, detectable with LD using the PIDS technique and Mie calculation. The best agreement with PCS measurements was observed for refractive indices of 1.33 and 1.44.

Fig. 56 Volume distribution of a 3.6 wt% Lipoid E80® (liposome) dispersion homogenised at 1000 bar and measured by LD with PIDS using different refractive indices for Mie calculation

However, liposomal dispersions showed weak diffraction intensities, which necessitated as much as 4 ml of the dispersions to be diluted to 150 ml. As liposomes represent hollow spheres, it is also unclear what value for their refractive index is valid. As is shown later, this causes problems when determining liposomes in the presence of oil droplets in emulsions (Section 4.3.2).
Ishii et al. [1990] and Chaturvedi et al. [1992] reported that submicron emulsions with 10% or 20% oil were less stable when lecithin was reduced below 1.2% w/v. Reduction of lecithin in commercially available 10% emulsions to 0.6% w/v has, however, already been realised. In commercial emulsions, lecithin:oil ratios of 0.12 (the older 10% emulsions), 0.06 (the newer 10% and 20% emulsions) and 0.04 (30% emulsions) are available. To examine the influence of this lecithin:oil ratio on emulsion properties, different oil contents of 5% up to 30% w/w emulsions were employed at fixed homogenisation pressure and amount of emulsifier. Fig. 57 shows the results for 10% and 20% w/w emulsions homogenised at 700 bar.

Fig. 57 Dependence of z-average diameter and polydispersity index on soya oil content of emulsions using 1.2 wt% Lipoid E80®: 20 wt% soya oil () and 10 wt% soya oil (O) [homogenised at 700 bar; n=5]

It is evident that particle size reduction and homogeneity are a function of number of cycles, as has been reported by Washington and Davis [1988] and Bock et al. [1994]. Contrary to their findings, however, particle size reduction and narrowing of distribution width continues until the last cycle, especially when a higher oil content is emulsified. Therefore, subsequent emulsions were homogenised for 11 cycles. Although the homogenised samples were kept at 30°C after each cycle, it can be assumed that on passing the interaction chamber they were heated by approx. 20°C before being cooled again by the heat exchange coil [Microfluidics International Corp., 1996]. For constant emulsifier concentration and homogenisation conditions, the dependence of z-average mean diameters on oil phase content yields a sigmoidal curve (Fig. 58), reported similarly by Ishii et al. [1990].

Fig. 58 Dependence of z-average diameter on soya oil content of emulsions using 1.2% w/w Lipoid E80® [homogenised for 11 cycles at 1000 bar; n=5]

Commercial and model 10% emulsions with reduced lecithin content show larger PCS z-average diameters (see Tab. 18) than with 1.2% lecithin. This could be caused either by larger emulsion droplets or by less emulsifier excess, which would, as liposomes, reduce the overall particle size measured. Fig. 59 illustrates the effect of reduced lecithin content on 10% model emulsion particle diameters. No difference in PI can, however, be seen.

Fig. 59 Dependence of z-average diameter and polydispersity index of 10 wt% soya oil emulsions on Lipoid E80® concentration: 0.6 wt% lecithin () and 1.2 wt% lecithin (O) [homogenised at 1000 bar; n=5]

Bock [1994] suggested that increasing emulsifier content in the oil phase counteracted droplet disruption by increasing dispersed phase viscosity. This could not be confirmed to have any influence on the 10% emulsion formulations investigated here.

3.2.3 Influence of Emulsifier Composition

Emulsions could not be stabilised using pure PC; Hansrani [1980] and Yamaguchi et al. [1995b] reported that certain ‘impurities’ were necessary to form stable emulsions. Emulsions from 1.2 wt% Lipoid® EPC (pure PC) indeed cracked immediately when 20% oil phase was incorporated, and showed measurable coarsening for 10% oil on 12 weeks’ storage. Surprisingly, the addition of only 0.02 wt% of sodium oleate yielded stable emulsions even with 20% oil (Tabs. 18 and 21), which showed similar particle size to emulsions containing Lipoid E80® (Fig. 61). The droplet interface coalescence experiments implicated, however, that addition of such small amounts of sodium oleate does not lead to such a pronounced effect on coalescence stability (Fig. 32). A clear effect of oleate could, however, be observed on the Zeta potential-pH-plots for Lipoid® EPC emulsions (Fig. 60), which may explain the remarkable increase in emulsion stability [Yamaguchi et al., 1995a].

Fig. 60 Influence of pH on Zeta potential for non-autoclaved 10% w/w soya oil emulsion stabilised with 1.2% w/w Lipoid EPC® (O) and 20% w/w emulsion with 1.2% w/w Lipoid EPC® + 0.02% w/w sodium oleate () [homogenised at 1000 bar]

Muchtar et al. [1991] reported that emulsions made with pure PC showed Zeta potential values of -30 mV, whereas stable emulsions from purified lecithin admixed with negatively-charged minor components exhibited values of -45 to -57 mV. The influence of pH and Zeta potential will be discussed in more detail in Section 3.2.5.
PC alone is obviously insufficient to stabilise the freshly-formed interface during high-pressure homogenisation. A water-soluble co-emulsifier like sodium oleate would locate at the interface faster and thus be more effective than PC. Fatty acids are either incorporated at the oil-water interface, possibly increasing repulsive surface charge by forming an integral part of the emulsifier film, or allow at least short-time stabilisation of the new interface until PC gradually becomes available and replaces the fatty acids. Sodium oleate added to emulsions prepared from Lipoid E80® also showed a slight reduction in z-average diameter (Fig. 61). However, a more pronounced effect was detectable for the early stages of homogenisation (cycles 1-3) where most of the uncovered, unstable oil-water interface was being created. Since the z-average diameter dropped faster for the admixed emulsions, a more rapid and effective stabilisation occurred for the systems containing the co-emulsifier. During cycles 4-11, where mainly disruption of remaining larger droplets occurs, the difference effectively disappears, as recoalescence becomes less important at this stage.

Fig. 61 Z-average diameter and polydispersity index of 20% w/w soya oil emulsions with 1.2% w/w Lipoid E80® (O) and additional 0.03% w/w sodium oleate () [homogenised at 1000 bar; n=5]

An even stronger effect is observed when Pluronic® F68 (dissolved in water) is used as the sole water-soluble emulsifier. It stabilised the newly-created interface immediately (Fig. 62). Within only one homogenisation cycle maximum dispersity is already reached and cannot be further reduced by repeated homogenisation. As Pluronic® F68 is a non-ionic emulsifier and produces sterically-stabilised emulsions [Eccleston, 1992], effects on Zeta potential were not assumed to be responsible for these findings. To allow application of high homogenisation pressure to improve dispersity of the emulsions and concomitantly avoid recoalescence during homogenisation, addition of a water-soluble compound such as fatty acid salts is very effective in improving the emulsification properties of lecithin. Depending on the duration of the pre-emulsification step, incipient hydrolysis of the lecithin could provide in situ production of fatty acids and, perhaps, also lyso-phospholipids (see Section 3.2.5.3). Since lyso-PC also reduces coalescence rates in the droplet/interface coalescence experiments (Fig. 30/ Tab. 11) and is more water-soluble than PC, the presence of both hydrolysis products may enhance stability against recoalescence during homogenisation. A marked effect of fatty acids on electrostatic repulsion during homogenisation is, however, only expected at alkaline pH, thus increasing their degree of ionisation (see also Section 3.2.5).

Fig. 62 Z-average diameter and polydispersity index of 20% w/w soya oil emulsions with 1.2% w/w Pluronic F68® [homogenised at 1000 bar; n=5]

Emulsions prepared from Lipoid E75® gave similar particle sizes as with Lipoid E80®. PCS results for homogenised lecithin dispersions also yielded comparable size values for the different lecithin types employed (Tab. 19).

3.2.4 Influence of Homogenisation Conditions

According to Washington and Davis [1988], Ishii et al. [1990] and Bock et al. [1994], the most important homogenisation parameters for controlling droplet size are homogenisation pressure, temperature and duration (number of cycles). From Washington and Davis [1988] and Bock [1994] it is known that higher temperature results in more effective disruption of droplets, but according to Bock et al. [1994] slightly broader particle size distributions and larger diameters for the largest droplets result when high temperature is used together with high homogenisation pressure for more than three cycles. This can be regarded as a typical case of overprocessed emulsions, where dispersion efficacy exceeds the stabilisation capability of the emulsifier. Stang and Schubert [1997] reported that stability against recoalescence could be considerably improved when critical droplet collisions are minimised by using a turbulent pipe flow adjacent to the dispersion zone. A conventional valve homogeniser was used, however, where cavitation was responsible for droplet disruption, and which yields larger particles for the same energy density compared with jet homogenisers like the Microfluidizer® [Schubert, 1997]. This might also account for their observation of generally broader size distributions.

Temperature was not varied in the following experiments to avoid irreproducible influence on chemical degradation (hydrolysis) of the lipids during homogenisation.

Fig. 63 gives the results for variation of microfluidization pressure for the example of model 20% emulsions. Higher emulsification pressure produces more rapid reduction of particle size and more homogeneous samples (as indicated by the more rapid reduction of PI values). This trend is consistent with the findings of Washington and Davis [1988], although these authors reported that homogenisation for only 6 cycles at approx. 700 bar yielded better dispersity than observed for 700 bar in Fig. 63. However, final particle sizes were also found to be the lowest for all samples where the highest pressure (1000 bar in our case) had been used.

Fig. 63 Influence of different homogenisation pressure on z-average diameter and polydispersity index of 20% w/w soya oil emulsions using 1.2% w/w Lipoid E80®: 1000 bar () and 700 bar (O); [n=5]

Evidently no overprocessing occurred for the samples at these conditions (1000 bar, 30°C). Judging from this result, emulsions should be homogenised at the highest pressure and for as many cycles as possible to achieve the smallest particle sizes and to disperse the emulsifier efficiently and thus reduce free lecithin content. However, to produce sufficiently small mean diameters with fewest proportion of droplets > 1 µm, just 3 - 5 cycles seem to suffice. To exploit more effectively the energy density applicable during homogenisation and yet avoid recoalescence, addition of a co-emulsifying agent like e.g. sodium oleate could improve the homogenisation results.
Tab. 18 and Tab. 19 summarise typical particle sizes (according to PCS and LD measurements) obtained for model emulsions and commercial formulations, as well as model liposome dispersions. These values will be referred to continually in the following sections.

Emulsion

Z-Average (nm)

Polydispersity
index

LD d50Vol (nm)

LD d90Vol (nm)

5%, 1000 bar, 1.2% LipoidE80®

151.8

0.124

270

840

Intralipid 10® (68460-51)

320.3

0.107

500

920

Lipovenoes 10 PLR®

348.1

0.098

450

790

Lipofundin 10% N®

347.1

0.161

450

800

Lipofundin MCT 10®

274.4

0.071

310

570

10%, 400 bar, 1.2% LipoidE80®

291.8

0.146

n.d.

n.d.

10%, 700 bar, 1.2% LipoidE80®

233.6

0.073

n.d.

n.d.

10%, 1000 bar, 0.6% LipoidE80®

239.6

0.057

330

750

10%, 1000 bar, 1.2% LipoidE80®

184.2

0.081

270

580

10%, 1000 bar, 1.2%LipoidEPC®

232.4

0.106

350

810

Intralipid 20® (85375-51)

365.4

0.165

480

880

Lipovenoes 20®

374.3

0.088

390

730

Lipofundin 20% N®

370.0

0.137

490

870

Lipofundin MCT 20®

282.5

0.073

350

590

20%, 700 bar, 1.2% LipoidE80®

401.6

0.377

n.d.

n.d.

20%, 1000 bar, 1.2% LipoidE80®

306.6

0.078

480

910

20%, 1000 bar, 1.2%LipoidE80®
0.03% sod.oleate

287.8

0.098

400

760

20%, 1000 bar, 1.2%LipoidEPC®
0.02% sod.oleate

309.6

0.119

410

710

20%, 1000 bar, 1.2%PluronicF68®

268.0

0.111

460

900

Intralipid 30® (87537-51)

513.5

0.127

570

960

30%, 1000 bar, 1.2%LipoidE80®

334.3

0.050

450

950

Tab. 18 Typical particle size distributions of commercial and model parenteral emulsions by PCS (intensity distribution) (left) and Laser Diffractometry [Mastersizer Micro, Fraunhofer Mode, 50% and 90% of volume distribution] (right)

For LD measurements volume distributions are typically reported as few larger particles can be detected more easily [Groves, 1984]. Similarly, presentation of diffraction data by Fraunhofer theory achieves better sensitivity towards larger droplets (> 5 µm) than using Mie theory [Kohlrausch and Steffens, 1997]. It can be seen that model emulsions produced at 1000 bar possess very small mean diameters, but are slightly broader in distribution than the commercial samples (d90 values). The smallest droplets are contained in the Lipofundin MCT® formulations, although the same amount of lecithin is contained in the LCT emulsions of the same manufacturer. As discussed above, the influence of oil phase content and emulsifier/oil ratio can also be deduced.

For preparation of solid lipid nanoparticles as ‘solidified’ model emulsion systems, several waxes and solid fatty acids were tested. The best homogenisation (melting) and stability behaviour was found for cetylpalmitate. However, higher amounts of emulsifier for stabilisation had to be used, yielding particle size distributions similar to those for parenteral emulsions although somewhat broader. Dilute SLN dispersions without emulsifier were also prepared by solidification of the fluid droplets by pouring the microfluidized stream directly into a beaker placed in an ice bath with stirring until solidification. Considerable amounts of large aggregates resulted, however, which had to be removed by centrifugation, and the remaining SLNs aggregated within a few days. The sizes reported in Tab. 20 are in agreement with data from the literature (e.g. Mueller et al. [1995]). Lecithin was found to be more effective as an emulsifying agent compared with Pluronic F68®, as aggregation occurred with dispersions prepared with the latter after prolonged storage in the refrigerator.

3.2.5 Influence of pH Adjustment and Autoclaving

Various emulsions and liposome dispersions were steam-sterilised for different times under pharmacopeial conditions (121°C, 2 bar). Particle size was monitored with PCS and Laser Diffractometry, together with microscopical evaluation, to follow droplet stability. Chemical degradation of the emulsions’ ingredients was determined by chromatographic analyses, which was also used to trace possible redistribution of lecithin components between aqueous and oil phase during autoclaving as had been reported by Groves and Herman [1993].

As had been proposed by Herman [1992], pH values in emulsions were measured after addition of increasing amounts of a saturated KCl-solution. By doing so, adsorption of ions onto the charged emulsion droplets (‘dispersion effect’/ see Lee et al. [1989]) should be reduced. A similar procedure is also used for pH determination of dextrose solutions according to USP23/NF18 [1995], where ions other than H+ and OH- are absent. pH-values obtained for addition of increasing proportions of saturated KCl-solution to 8 ml of commercial emulsion samples are shown in Fig. 64. The pH values drop markedly by approx. one unit upon addition of KCl-solution, although KCl-solution was presumed to possess neutral or slightly acidic pH of 6-7. pH reached a plateau after addition of 500 µl solution to the original samples. However, Herman and Groves [1993] who added 50 µl to 2 ml of emulsion samples, reported that only half of this amount was necessary. Bock [1994], however, observed that electrode equilibrium is reached more slowly when electrolyte solution is added. Thus it was decided to measure pH in the subnatant aqueous phase obtained after centrifugation for 24 h at approx. 13000 g. pH measured in this way did not, however, differ from pH determined by direct measurement in the emulsions, showing that the dispersion effect does not occur. Values given here were therefore obtained from the untreated emulsions.

Fig. 64 Influence of addition of saturated KCl-solution on pH of commercial parenteral emulsions: Lipofundin 20% N® (O), Intralipid 10® (), Lipovenoes LCT 20® ()

3.2.5.1 Zeta Potential

Spooner et al. [1990] reported an apparent pKa for oleic acid of 7.4 - 7.5 determined by NMR spectroscopy for an equal distribution of the fatty acid between phospholipid bilayers and lecithin on the surface of emulsion droplets. Similarly, Spector [1975] stated that the pKa of long-chain fatty acids ranged in a more ‘hydrophobic’ environment from about 4.7 - 5.0, although 6 - 8 had also been reported by other authors. Villalaín and Gómez-Fernández [1992] determined the apparent pKa of palmitic acid in multilamellar vesicles to be about 8.7 using FT-IR spectroscopy. Accordingly, Herman [1992] argued that the influence of charged fatty acids located at the oil-water interface on emulsion stabilisation should be negligible. To estimate the influence of pH on free fatty acid ionisation, Fig. 65 shows dissociation curves for the example of oleic acid according to its higher and its possible lower pKas. Even with pKa = 7.5, considerable ionisation occurs for pH values between 9 and 11. Although electrostatic repulsion owing to free fatty acids might not be enhanced at neutral pH (as in the final product), a stabilising effect might be postulated, nevertheless, for the emulsification and autoclaving steps where alkaline pH values exist. It can also be seen from Fig. 65 that emulsions prepared from either pure PC or non-ionic emulsifier Pluronic F68® exhibit low negative or even slightly positive Zeta potential values under acidic conditions and become shifted to more negative values with increasing pH. Nevertheless, the change of Zeta potential with pH shift from neutral to alkaline (where ionisation of fatty acids would be expected to become more important) is not very pronounced. Little contribution of Zeta potential to droplet stabilisation would, therefore, be expected in the case of pure PC.

Fig. 65 Influence of pH on (A) dissociation of oleic acid and on (B) Zeta potential for non-autoclaved 10% w/w soya oil emulsions stabilised with 1.2% w/w Lipoid EPC® (O) and 1.2% w/w Pluronic F68® () [homogenised at 1000 bar]

Only small amounts of free fatty acid can be present before autoclaving, since otherwise they would increase Zeta potential considerably.
Zeta potentials of Lipoid E80®-stabilised emulsions were found to increase by some 10 - 20 mV during autoclaving, accompanied by a parallel drop in emulsion pH. Emulsions autoclaved without addition of NaOH to adjust pH to alkaline showed cracking and shifting of pH towards 3 - 4, although these emulsions finally exhibited the highest Zeta potential values. The example of a 20% emulsion (pH 6.5 before autoclaving and pH 3.5 after autoclaving for 30 min) is depicted in Fig. 66. The autoclaved, cracked emulsion possesses nominally higher Zeta potential at pH > 5, but < pH 5 both emulsions have similar Zeta potentials. Taking pKa-values of 11.25 (DLPE, NH3), <1 (DPPC, PO4), about 6 (DPPS, COOH) and 3.87 (egg-PA, PO4), (Marsh [1990]), about 10% of PS carboxylic groups are still dissociated at pH 5, whereas long-chain fatty acids would be expected to be almost entirely undissociated (Fig. 65, pKa = 7.5). Free fatty acids possibly produced by hydrolysis during autoclaving are therefore not able to increase Zeta potential under acidic conditions and fail to enhance emulsion stability. pH adjustment to alkaline values has not only to be regarded as a means to ‘pre-adjust’ final emulsions to blood pH, but is rather a prerequisite to allow autoclaving of parenteral emulsions [Bock, 1994].

Addition of sodium oleate to emulsions stabilised with pure PC (Lipoid EPC®) influenced Zeta potential in a similar fashion (Fig. 60). Also, only the oleate-admixed emulsions could be sterilised without cracking, and not those with pure PC alone, although in all cases pH had been adjusted to > 9.0 before autoclaving. Stabilisation owing to increased electrostatic repulsion thus appears only to be effective at Zeta potentials > approx. -50 mV for emulsions containing PC.

Fig. 66 Influence of pH on Zeta potential for 20% w/w soya oil emulsion stabilised with 1.2% w/w Lipoid E80® non-autoclaved (O) and after autoclaving for 30 min at 121°C without prior pH adjustment () [homogenised at 1000 bar]

However, even for pH values < 5, oleate-admixed emulsions showed Zeta potentials approx. 10 mV higher than those with pure PC (Fig. 60). It is also evident from the high Zeta potentials of broken emulsions that high Zeta potential alone cannot be regarded as an indicator for stability. However, as a pronounced effect could be observed for droplet stability during heat stress, it is important to control pH and resulting Zeta potential values before autoclaving. Bock [1994] recommended adjusting of emulsions to alkaline pH prior to homogenisation instead of prior to autoclaving, thus resulting in larger Zeta potentials with less variability in final pH.
In the case of commercial emulsions similar Zeta potentials and dependency on pH were observed. Both Lipofundin MCT® 10% (5 months prior to expiration) containing oleate admixture, as well as Intralipid 20® (6 months prior to expiration) containing no additional oleate, showed Zeta potentials of approx. -50 mV at pH 7. Higher values occurred at stronger alkaline pH, which is not seen with the Lipoid EPC®-emulsions where Zeta potentials reach a plateau within the alkaline pH range (Fig. 65). Since the Lipoid E80®-emulsions show a similar pH-dependency before and after autoclaving (Fig. 67), higher Zeta potentials of the commercial emulsions can be attributed to additional minor lecithin components which were absent in the emulsions containing pure PC. The Point of Zero Charge (PZC) of the emulsions lay around or below pH 3, which is in agreement with reports by Dawes and Groves [1978] and Lucks [1993] for Intralipid® formulations.
To verify that Zeta potentials measured at 25°C reflect accurately those values existing at elevated temperature during production, Zeta potential measurements were also carried out on Intralipid 20® at various temperatures using sample dilutions of pH 8 and pH 9. Similar to the reports of Yamaguchi et al. [1995b], only slight changes of the Zeta potentials were observed (Fig. 68). This is an additional indication of the validity of the importance of electrostatic repulsion for autoclaving stability of these emulsions at high temperature.

Fig. 67 Influence of pH on Zeta potential for Intralipid 20® [batch 85375-51] (), Lipofundin MCT® 10% [batch 6023A81] () and non-autoclaved 20% w/w model emulsion stabilised with 1.2% w/w Lipoid E80® (O) [homogenised at 1000 bar]

Fig. 68 Influence of temperature on Zeta potential for Intralipid® 20 [batch 85375-51] at pH 8 (O) and pH 9 ()

Model emulsions consisting of liquid paraffin as the dispersed phase instead of soya oil, showed a similar drop in pH and increase in Zeta potential on autoclaving. This effect owes therefore more to the emulsifier than to hydrolysis of the dispersed phase.

Essentially, these Zeta potential values confirm data reported for similar emulsions by Washington and Davis [1987], Washington et al. [1989], Chaturvedi et al. [1992] and Yamaguchi et al. [1995a]. They contradict reports by Muchtar et al. [1991], who claimed that emulsions prepared with purified lecithin, enriched in negatively-charged phospholipids, experienced neither changes in Zeta potentials nor pH values when different autoclaving times were used. The production of hydrolysis products of lecithin during autoclaving would be expected to enhance surface charge of oil droplets, but also to lower pH leading to diminished proportion of ionised species (and thus lower Zeta potential), as reported by Washington and Davis [1987] by adding oleic acid to parenteral emulsions. As no difference between Zeta potentials of emulsions containing 0.6% or 1.2% lecithin could be observed, reports by Ishii et al. [1990] could not be confirmed. These authors found Zeta potential values to be maximally negative at 1.2% lecithin content, but did not state the pH of their emulsions. It appears therefore that Zeta potential and with it also droplet stability during autoclaving are not affected by reduction of excess emulsifier, although the latter could theoretically serve as a ‘reservoir’ of hydrolyseable phospholipids and thus as a source of free fatty acids. Older reports on emulsion Zeta potential refer to the outdated Intralipid 10® formulation containing excess lecithin. This might, however, also have affected their admixing stability, since charged lipids might be able to buffer ion admixing. However, Rubino [1990] stressed that emulsions to which sodium oleate had been added or ‘aged’ commercial emulsions (which contain more hydrolysed lecithin) exhibited increased flocculation upon the addition of Ca2+-ions, whereas addition of sodium-PA gave best stabilisation against flocculation. Also Washington et al. [1990] used determination of critical flocculation concentrations of added Ca2+ to determine ‘stability’ of emulsions. They reported that increase in pH markedly increased stability of Intralipid® against Ca2+ addition, which was even further improved when glucose solution was contained.

3.2.5.2 Physical Stability

Tab. 21 shows to which degree particle sizes of different emulsions were affected by autoclaving. According to reports in the literature [Lee and Groves, 1981 / Groves and Herman, 1993], autoclaving of model emulsions was expected to yield only a slight decrease in PCS z-average diameters, if at all. The former finding reflects the disappearance of few larger particles from the measurement range owing to aggregation or coalescence. However, in these experiments, particle size remained practically unchanged for the pH-adjusted emulsions, whereas the emulsions whose pH had not been adjusted before autoclaving showed marked coarsening (see 3.2.5.1). The addition of sodium oleate clearly increased the autoclaving-stability of emulsions stabilised with pure Lipoid EPC®.

Emulsion

Z-Average (nm)

Polydispersity
index

LD d50Vol (nm)

LD d90Vol (nm)

10%, 1000 bar, 0.6%LipoidE80®

239.6

0.057

330

750

10%, 1000 bar, 0.6%LipoidE80®

245.1

0.035

330

800

10%, 1000 bar, 1.2%LipoidE80®

184.2

0.081

270

580

10%, 1000 bar, 1.2%LipoidE80®

186.7

0.075

n.d.

n.d.

10%, 700 bar, 1.2%LipoidE80®

233.6

0.073

n.d.

n.d.

10%, 700 bar, 1.2%LipoidE80®

236.4

0.079

n.d.

n.d.

10%, 1000 bar, 1.2%Lipo.EPC®

232.4

0.106

340

780

10%, 1000 bar, 1.2%Lipo.EPC®

1235.9

0.200

1360

2390

20%, 1000 bar, 1.2%LipoidEPC®
0.02% sod.oleate

309.6

0.119

410

710

20%, 1000 bar, 1.2%LipoidEPC®
0.02% sod.oleate

305.9

0.080

410

720

20%, 1000 bar, 1.2%LipoidE80®

256.3

0.056

400

745

20%, 1000 bar, 1.2%LipoidE80®

263.4

0.060

400

730

20%, 1000 bar, 1.2%LipoidE80®, no pH adjustmnt.

1387.0

0.250

n.d.

n.d.

Intralipid 20® [batch 85375-51]

342.0

0.134

480

880

Intralipid 20® [batch 85375-51]

351.1

0.095

480

890

30%, 1000 bar, 1.2%LipoidE80®

334.3

0.050

450

950

30%, 1000 bar, 1.2%LipoidE80®

344.9

0.057

420

810

Tab. 21 Typical particle size distributions of commercial and model parenteral emulsions by PCS (intensity distribution) (left) and Laser Diffractometry [Mastersizer Micro, Fraunhofer Mode, 50% and 90% of volume distribution] (right) before (plain) and after (shaded) autoclaving for 15 min at 121°C

Repeated autoclaving of commercial emulsions also was not found to lead to major changes in their particle size distributions.

Although liposomal dispersions generally showed broad size distributions, PCS results suggest, that liposomal structures and sizes were maintained during autoclaving up to 5 h duration. This was not seen when pH had not been adjusted to alkaline values before autoclaving and accordingly dropped to acidic values. Liposome size was similarly stable for pure egg-PC (Lipoid EPC®) and the less purified blend (Lipoid E80®) and therefore independent of emulsifier composition.

Sample

Z-Average (nm)

Polydispersity
index

1.2% Lipoid E80®, 400 bar

156.5

0.546

1.2% Lipoid E80®, 400 bar

142.2

0.499

1.2% Lipoid E80®, 700 bar

115.4

0.752

1.2% Lipoid E80®, 700 bar (5 h, pH 9)

100.4

0.682

1.2% Lipoid E80®, 700 bar (5 h, pH 6)

376.5

0.265

4.0% Lipoid EPC®, 1000 bar

75.0

0.567

4.0% Lipoid EPC®, 1000 bar

75.5

0.600

4.0% Lipoid E80®, 1000 bar

74.1

0.557

4.0% Lipoid E80®, 1000 bar

77.8

0.612

Tab. 22 Particle size distributions of microfluidized liposomal dispersions of egg lecithin by PCS (intensity distribution) before (plain) and after (shaded) autoclaving for 15 min at 121°C or as indicated

3.2.5.3 Chemical Composition

2-propanol : n-hexane (1:1 v/v) was found to be the best solvent for the lyophilised combined triglycerides, fatty acids, polar and apolar lipids. In some cases, dissolution could be accelerated by short-time gentle heating of the samples. No filtering or pre-purification was needed, and after 500 injections no backpressure increase was observed during HPLC analysis. The relative standard deviations between three successive injections were all approx. 1.5% for PC or PE and about 3% for the low-response signals from LPC. These values are comparable to the reports of Sotirhos et al. [1986b], except LPC which was reported to be detected with a relative error of about 10%.
Despite the relatively slow pump rate, no serious tailing of the solvent front was observed, as reported by Herman [1992]. The elution of triglycerides (together with the solvent front) and free fatty acids, approximately 1-2 minutes later, could therefore be observed. This could be confirmed by spiking samples with oleic and myristic acid. The first phospholipid of interest, PE, eluted well-separated from the oil-related lipids after about 15 min. LPC eluted at about 47 min, after which the gradient was returned to the initial conditions. Although this procedure required relatively long analysis times, reliable separation was achieved. Higher flow rates or more hydrophilic gradient conditions could probably have reduced run times or improved peak shape [Herman, 1992]. Quantitation of egg lecithin samples is, however, very complex and tedious, and the results reported here are sufficient for the purpose of characterisation of the samples under investigation. For more detailed studies and reviews on different solvents and detection techniques the reader is referred to McCluer et al. [1986], Herman [1992], Beare-Rogers et al. [1992] and Balazs et al. [1996].
Consistent with the reports of Herman [1992], Herman and Groves [1992] and Grit et al. [1989], hydrolysis of PC in both model and commercial emulsions occurs according to first order kinetics (Fig. 69). PC in liposomal dispersions undergoes faster degradation than in the 20% model emulsions. Contrary to Herman and Groves’ suggestion, any catalyst effects of a buffered aqueous medium can be ruled out.

Fig. 69 Semilogarithmic plot for PC degradation during autoclaving of Intralipid 20® [batch 70178-51] (), 20% w/w model emulsion stabilised with 1.2% w/w Lipoid E80® without pH adjustment [pH 6.6] (O) and 1.2% w/w Lipoid E80® dispersion [pH6.2] ()

A possible explanation is that the PC located in the lamellar bilayers was more easily hydrolysed by the surrounding water than in the more lipophilic environment at the emulsions’ oil-water-interface. As a result, the higher the initial content of excess lecithin in an emulsion the greater should be the lyso-PC content in the final product. Herman [1992] and Herman and Groves [1992] give various values for Arrhenius activation energies (Ea) for PC and PE hydrolysis in their model emulsions. They, however, found that phospholipid hydrolysis was independent of the purity of the oil used as the dispersed phase. Intralipid 20® shows less degradation of PC when re-autoclaved compared with the model dispersions being autoclaved for the first time. Consequentially, its pH also decreased less (Fig. 70). Confirming the report of Herman [1992], the drop in pH during the first 15 min interval is greater than can be related to mere hydrolysis of the phospholipids. Yet, the extent of decrease (2-3 units) cannot clearly be related to the pH adjusted prior to autoclaving (Fig. 70). Bock [1994] observed that even non-autoclaved emulsions which had been pH-adjusted after homogenisation showed a drop in pH of about 1-2 units within 24 h, whereas those whose pH had been adjusted prior to homogenisation did not change pH remarkably. He attributed these findings to neutralisation reactions or incipient triglyceride hydrolysis which had been accelerated by the homogenisation process. This might also explain the initial pH drop in Fig. 70, as all systems were examined immediately after preparation.

Fig. 70 Drop in pH during different autoclaving intervals: Intralipid 20® [batch 70178-51] (), 20% w/w model emulsion stabilised with 1.2% w/w Lipoid E80® without prior pH adjustment (¨), adjusted to pH 9 (O) and pH 10 (), as well as 1.2% w/w Lipoid E80® dispersion without prior pH adjustment ()

To maintain stability of the emulsions it appears, therefore, that pH adjustment before autoclaving must be controlled carefully to avoid a subsequent shift during autoclaving to lower pHs. This can be achieved by adjusting pH of the aqueous phase before homogenisation (as suggested by Bock [1994]) or by adjusting pH to about 10 after homogenisation, taking into account the subsequent drop of almost 3 pH units to be expected (Fig. 70). Many authors reporting difficulties with autoclaving stability either did not report pH or probably oversaw these pH changes.

The chromatograms of 20% model emulsions depicted in Fig. 71 show that with decreasing intensity of PE (Fig. 71, 2) and PC signals (Fig. 71, 4), free fatty acids, lyso-PC and also a third signal at about 22 min are detected. From the retention times reported by Sotirhos et al. [1986b], this signal could be related to the formation of lyso-PE. The values observed for first-order rate constants were in agreement with the one which could be derived from the cited reports of Herman on PC hydrolysis after application of high-temperature stress towards model emulsions (Tab. 23).

Fig. 71 HPLC chromatograms of native 20% w/w model emulsions stabilised with 1.2% w/w Lipoid E80® [pH 6.6] () and after autoclaving for [A] 15 min (-) and [B] 5 h (-) without prior pH adjustment:

1-FFA 2-PE 3-LPE 4-PC 5-LPC

Fig. 71 and Tab. 23 also indicate that PE is in most cases hydrolysed faster than PC, whereas Herman and Groves [1992] reported almost equal hydrolysis rates for both compounds. However, hydrolysis rates are clearly found to be pH-dependent, and different pHs used in their emulsions might account for these different outcomes. PE yielded increased hydrolysis rates at alkaline pH, whereas PC appeared to be most stable at pH 9. This was surprising, since Håkansson [1966] had predicted hydrolysis to appear at the slowest rates at pH of 6.5. Since the detection of PC was in agreement with the data of Herman and Groves, this inexplicable deviation can only in part be caused by the method of detection applied (UV vs. FID used by Herman and Groves).

Fig. 72 Semilogarithmic plots for (A) PC and (B) PE degradation during autoclaving of 20% w/w model emulsions stabilised with 1.2% w/w Lipoid E80®, adjusted to pH 9.0 (), pH 10.0 () and without prior pH adjustment [pH6.6] (O)

Tab. 23 summarises the first-order hydrolysis rates observed for PC and PE:

Sample
(pH before autoclaving)

kobs [1/h]

 

PC

PE

1.2% Lipoid E80® liposomal dispersion (6.2)

0.1241

0.1023

20% Model Emulsion (6.6)

0.0651

0.0875

20% Model Emulsion (9.0)

0.0446

0.1386

20% Model Emulsion (10.0)

0.0497

0.1383

Intralipid 20®
[batch 70178-51] (7.52)

0.0056

0.0338

20% Emulsion (8.5) *

0.0501

0.0446

Tab. 23 First-order hydrolysis rate constants for PC and PE in 20% emulsions autoclaved at 121°C as determined by HPLC (* calculated from Herman [1992])

Emulsifier degradation during only 15-20 minutes’ autoclaving proceeds slowly, making it difficult to obtain reliable degradation data from change in PC or PE content. Therefore, corresponding LPC values as determined by HPLC analysis are depicted in Fig. 73.

Fig. 73 Semilogarithmic plot for increase of LPC content during autoclaving: 20% w/w model emulsion stabilised with 1.2% w/w Lipoid E80® without prior pH adjustment [pH6.6] (O), adjusted to pH 9.0 prior to autoclaving (), and non-adjusted 1.2% w/w Lipoid E80® dispersion [pH6.2] ()

Despite the limitations regarding LPC quantitation by UV-absorbance (see Section 3.1.2.1), substantially the same dependency between hydrolysis and pre-autoclaving sample pH are seen. However, LPC content does not increase parallel to PC degradation and deviates from first-order kinetics. Herman [1992] and Herman and Groves [1992] suggested that ongoing hydrolysis of LPC to GPC is an explanation for this behaviour, and Chaturvedi et al. [1992] stated that direct hydrolysis of PC to GPC occurred especially at alkaline pH. Indeed, after prolonged autoclaving two new peak shoulders (indicated by the arrows in Fig. 74) emerge. Herman and Groves [1992] and Herman [1992] reported more detailed data on GPC/GPE content.

Fig. 74 HPLC chromatograms of native 1.2% w/w Lipoid E80® dispersion [pH 6.2] () and after autoclaving for 5 h (-) without prior pH adjustment:

1-FFA 2-PE 3-LPE 4-PC 5-LPC

HPTLC measurements gave LPC contents ranging from about 3 to almost 19% (mol:mol total phospholipid) in various emulsion samples (Tab. 24). Phospholipid hydrolysis had, therefore, obviously begun before autoclaving, owing to the initial dispersion stages where elevated temperatures had been used. It may be assumed, therefore, that addition of NaOH at this stage may also have produced dissociation of free fatty acids before the homogenisation step. As in commercial products Na-oleate is added (approx. 610-4 to 110-3 mol/l) to adjust the pH of the aqueous phase before homogenisation, a co-stabilising effect of free fatty acids may be possible by phospholipid hydrolysis before homogenisation.

Sample (pH before autoclaving)

LPC (mol%) before

LPC (mol%) after

20% Model Emulsion (6.6)

3.2

5.8

Intralipid 20® *
[batch 70178-51] (7.52)

18.6

19.7

Tab. 24 Results for HPTLC determination of LPC content of non-pH- adjusted emulsion before and after autoclaving for 15 min at 121°C (* analysed 2 months after expiry)

Comparison with Intralipid 20® (Tab. 24) also confirms, that LPC content rises further during storage, producing concentrations well above 10% around the time of expiry, which is in agreement with earlier reports from Muehlebach et al. [1987] and Herman [1992] who had found up to about 10% LPC in emulsions still within their shelf-life by HPTLC and HPLC analysis. This supports the prediction of Hansrani [1980] and Washington and Davis [1987] that stability was better in aged emulsions than in freshly produced samples.
Using HPLC analysis it was also possible to reveal differences in emulsifier composition between different commercial products and compare their composition with the model emulsions. Model emulsions (adjusted to pH 9.0) showed only slightly detectable degradation when analysed before and immediately after autoclaving (Fig. 75a). Identical chromatograms can be seen, and only a slight increase in free fatty acid peaks together with minor signals for LPC and LPE are detectable. From Fig. 75b it is evident that Lipofundin 10%N® and Lipovenoes LCT 20® possess almost identical phospholipid composition, with a PC:PE ratio similar to that for the model emulsions. These emulsions were analysed shortly before their expiry, and clearly contain pronounced amounts of free fatty acids and lyso-phospholipids, whereas both PC and PE content are diminished compared with the model emulsions. This observation emphasises again that hydrolysis of the phospholipids continued during storage. In Fig. 75c, Lipofundin MCT 10%® is shown to contain substantially more hydrolysis products than Intralipid 10®. The total amount of phospholipids in Lipofundin MCT 10%® is, however, nominally twice the amount used in Intralipid 10®. Nevertheless, PE is present in equal amounts in both emulsions, indicating that the Intralipid® formulation possessed a higher PE:PC ratio, as could also be seen from Tab. 25, and had similarly been reported by Muehlebach et al. [1987].

(A) 20% w/w model emulsions stabilised with 1.2% w/w Lipoid E80® and adjusted to
pH 9.0, before (
) and after autoclaving (-) for 15 min at 121°C
(B) Lipofundin 10% N® [1 month prior to expiry] (-) and Lipovenoes LCT 20®
[1 month prior to expiry] (
)

(C) Lipofundin MCT 10%® [1 month prior to expiry] () and Intralipid 10®
[batch 87535-51, 12 months prior to expiry] (-)
(D) Lipofundin MCT 20%® [1 month prior to expiry] (-) and Intralipid 30®
[batch87537-51, 12 months prior to expiry] ()

Fig. 75 HPLC chromatograms of commercial and model parenteral emulsions: 1-FFA 2-PE 3-LPE 4-PC 5-LPC

A similar result is depicted in Fig. 75d, showing the phospholipid compositions of Lipofundin MCT 20%® and Intralipid 30%® which are both formulated with 1.2% lecithin. Intralipid® also in this case shows more intense PE and LPE peaks compared with the Lipofundin® emulsion. Intralipid 30® [batch 88555-71] contained less PC but more PE than found in Intralipid 10® and 20®, which was consistent with reports by Férézou et al. [1994]. These authors found increased contents of PE and PI at the expense of PC in Intralipid 30®. However, duplicate analysis of the three emulsion types from batches one year ahead of their expiry did not confirm such differences between the three formulations and suggests that the former finding was simply batch-to-batch variations of the emulsifier. From Fig. 75d it can, however, be seen that Intralipid 30® contains more free fatty acids than usually observed for 10% or 20% emulsions. As Intralipid 30® samples had been analysed well before expiry, either more hydrolysis of lecithin owing to heat stress had taken place, or, more likely, more free fatty acids had been introduced by the increased proportion of the oil phase. This may occur either by the inherent free fatty acid content of the soya oil, or by hydrolysis of the latter during production and storage. For emulsions stabilised with Pluronic F68®, homogenised and autoclaved under the same conditions, no substantial amounts of free fatty acids were released during autoclaving (Fig. 76). Thus the oil phase is not the source of fatty acids. Free fatty acid release under autoclaving conditions is therefore a result of degradation of phospholipids.

Fig. 76 HPLC chromatograms of native 20% w/w emulsion stabilised with 1.2% w/w Pluronic F68® [pH 5.5] () and after autoclaving for 20 min at 121°C without prior pH adjustment (-) and after adjustment to pH 9.0 (N) [enlarged section showing typical free fatty acid retention times]

Fig. 77 shows the same section of the chromatograms of a 20% w/w emulsion stabilised with Lipoid EPC® and 0.02% w/w sodium oleate before and after autoclaving. Due to the addition of sodium oleate, a pH of 9.26 was observed and therefore, no additional NaOH was used to control the pH. It is evident that hydrolysis led to formation of further free fatty acids, and that the initial oleate content is small compared with the usual content of hydrolysis-derived FFAs (compare with Fig. 75).

Fig. 77 HPLC chromatograms of native 20% w/w emulsion stabilised with 1.2% w/w Lipoid EPC® and 0.02% w/w sodium oleate before [pH = 9.26] () and after autoclaving for 15 min at 121°C (-) [enlarged section showing typical free fatty acid retention times]

Since the composition of the lecithin in the emulsions, especially the ratio of PC:PE, are regarded as important for emulsion stability [Hansrani, 1980 / Groves and Herman, 1993], PC and PE contents of various commercial and autoclaved model emulsions determined by HPLC analysis are summarised in Tab. 25. The PC and PE composition of the model emulsions possess highest PC:PE ratios but were similar to the Lipofundin® and Lipovenoes® formulations. In contrast, Intralipid® formulations in general possess lower PC:PE ratios. However, molar contents of total phospholipids in Intralipid® batches sometimes ranged at the upper end or even exceeded the amount theoretically contained. This can again be explained by differences in the fatty acid substitution of the phospholipids, which influence also the molar mass. The variations in lecithin composition observed in Tab. 25 are, however, not relevant for emulsion autoclaving stability.

Emulsion

PC (mol/l)

PE (mol/l)

PC+PE (mol/l)

PC:PE ratio

Intralipid 10® 1
(batch 68460-51)

5.53410-3

1.40510-3

6.93910-3

3.9 : 1

Intralipid 10® 12
(batch 87535-51)

6.31210-3

1.76110-3

8.07310-3

3.6 : 1

Lipovenoes 10 PLR®

5.98010-3

8.03010-4

6.78310-3

7.4 : 1

Lipofundin MCT10®

1.16310-2

1.51410-3

1.31410-2

7.7 : 1

10%, 1000 bar,
1.2% LipoidE80®

1.08110-2

1.21510-3

1.20210-2

8.9 : 1

Intralipid 20® 1
(batch 70178-51)

1.02610-2

2.86110-3

1.31210-2

3.6 : 1

Intralipid 20® 12
(batch 85375-51)

1.16910-2

3.18810-3

1.48810-2

3.7 : 1

Lipovenoes 20®

1.12710-2

1.67710-3

1.29510-2

6.7 : 1

Lipofundin 20% N®

9.36310-3

1.66110-3

1.10210-2

5.6 : 1

Lipofundin MCT20®

9.90310-3

1.87710-3

1.17810-2

5.3 : 1

20%, 1000 bar,
1.2% LipoidE80®

1.27710-2

1.50110-3

1.42710-2

8.5 : 1

20%, 1000 bar, 1.2%LipoidEPC®,
0.02% sod.oleate

1.23010-2

-

1.23010-2

-

Intralipid 30®
(batch 88555-71)

1.21210-2

3.76610-3

1.58910-2

3.2 : 1

Intralipid 30®
(batch 87537-51)

1.32510-2

3.67310-3

1.69210-2

3.6 : 1

30%, 1000 bar, 1.2%LipoidE80®

1.37610-2

1.67110-3

1.54310-2

8.2 : 1

Tab. 25 PC and PE content of commercial and model emulsions as determined by HPLC analysis
(1-analysed 1 month prior to expiry, 12-analysed 12 months prior to expiry)

3.2.5.4 Emulsifier Distribution

Heat stressing of lecithin-stabilised emulsions has been reported to lead to irreversible redistribution of emulsifier compounds within oil and aqueous compartments [Herman, 1992 / Groves and Herman, 1993]. PC and PE were reported to move from the aqueous continuum towards the dispersed oil phase. Emulsion oil and aqueous compartments were separated by centrifugation in a swing-out rotor at 13000g for approx. 18h (described in more detail in Chapter 4) which allowed phase separation without destruction of the oil droplets. As shown in Fig. 78, the aqueous phase of centrifuged Intralipid 30® contained measurable amounts of phospholipids and, indicated by the intense, broad peak around the solvent front, a large proportion of triglycerides. This is in agreement with the reports of Rotenberg et al. [1991], Westesen and Wehler [1992] and Férézou et al. [1994], who reported many oil droplets in the subnatant. Owing to enhanced Brownian motion, acceleration was too low to separate the smallest oil droplets during centrifugation.

Most of the hydrolysis products (LPC, LPE and free fatty acids) were still associated with the oil droplets and had not moved to the aqueous compartment to form micelles (see Tab. 26). Westesen and Wehler [1992] did not find large amounts of lyso-phospholipids in the aqueous phase either, which suggests that hydrolysis products remain within the emulsifier film, enhancing electrostatic repulsion as well as influencing film properties (see Section 3.1.1.1). Also Herman [1992] reported LPC content of the oil to the aqueous phase of about 7:1, and that approximately twice the amount of LPC was located in the oil phase than could be accounted for according to its initial proportion of the total phospholipid content. He also observed a relocation of PC from the aqueous to the oil phase owing to heat-stress which occurred faster than for PE. Thus, PC:PE ratio of the oil phase increased during autoclaving. This has inter alia been suggested to be the cause of reversible cubic phase formation of the lecithin upon thermal stress, possibly responsible for enhanced autoclaving stability [Groves and Herman, 1993].

(A) Intralipid 30® [batch 88555-71, 15 months prior to expiry] cream layer () and aqueous subnatant (-) after centrifugation
(B) Intralipid 10® [batch 68460-51, 1 month prior to expiry] cream layer () and aqueous subnatant (-) after centrifugation

(C) Lipovenoes LCT 20® [1 month prior to expiry] cream layer () and
aqueous subnatant (
-) after centrifugation
(D) Lipovenoes LCT 10 PLR® [6 months after expiry] cream layer () and
aqueous subnatant (
-) after centrifugation

Fig. 78 HPLC chromatograms of creamed oil phase and aqueous subnatant of a selection of commercial parenteral emulsions:

1-FFA 2-PE 3-LPE 4-PC 5-LPC

Fig. 79 shows that also for model emulsions a fraction of the triglycerides and all types of phospholipids were present in the aqueous phase after centrifugation. Compared with commercial emulsions (Fig. 78 and Tab. 26), however, less phospholipid is contained in the aqueous phase; it is present mainly in association with the oil phase.

Fig. 79 HPLC chromatograms of a 30% model emulsion stabilised with 1.2 wt% Lipoid E80®, adjusted to pH 9.0:
(A) Aqueous subnatant prior to autoclaving (-) and after autoclaving for 20 min at 121°C (N) after centrifugation (with enlarged inserts) and (B) respective cream layer before (-) and after autoclaving (N)

1-FFA 2-PE 3-PC 4-SPM 5-LPC

However, a clear reduction in PE and PC is observed for the aqueous compartment as a consequence of autoclaving. LPC and FFA contents are, surprisingly, not changed by phospholipid hydrolysis during heat stress. PC and PE content in the cream layer also decreases upon autoclaving, however the ratio of the intact phospholipids in the cream layer to the water phase is greatly increased (Tab. 26). This is accompanied by a notable increase in free fatty acid and LPC peaks. It appears that hydrolysis products were distributed in favour of the oil compartment, and those formerly present in the water were largely relocated to the oil-water-interface. For those phospholipids associated to the O/W-interface, their hydrolysis products remain at the interface. This would counteract the formation of micelles from LPC or FFAs in the aqueous phase. Kumar et al. [1989] and Herman [1992] suspected similarly that incorporation of these compounds into vesicles and emulsion droplets would account for the reduced haemolytic toxicity found for these systems. SPM is also reduced in the aqueous subnatant after autoclaving, yet no corresponding changes are seen for the cream layer.

Fig. 80 shows how for 10% w/w model emulsions produced at 1000 bar a redistribution of phospholipids can be observed.

Fig. 80 HPLC chromatograms of a 10% model emulsion stabilised with 1.2 wt% Lipoid E80®, adjusted to pH 9.0:
(A) Aqueous subnatant prior to autoclaving (-) and after autoclaving for 20 min at 121°C (N) after centrifugation (with enlarged inserts) and (B) respective cream layer before (-) and after autoclaving (N)

1-FFA 2-PE 3-PC 4-SPM 5-LPC

The PC and PE peaks are clearly diminished in the aqueous compartment after autoclaving, whereas LPC and FFAs remain unchanged. Surprisingly, PC is enhanced in the cream layer and PE unchanged, although markedly increased hydrolysis peaks are seen. This also indicates a redistribution of phospholipids from the aqueous to the oil compartment, as was reported by Herman [1992]. Compared with Fig. 79, the total phospholipid concentration of the aqueous compartment of the 10% formulations was greater than for the 30% emulsion, supporting the assumption that excess lecithin in the 10% formulation is present in the aqueous phase.

It was, however, not possible to confirm the reports of Herman [1992] and Groves and Herman [1993] that PC redistributes preferentially from the aqueous to the oil compartment upon heating, thus producing a higher PC:PE ratio in the cream layer of their model emulsions. In contrast to their findings, no clear trend for redistribution of distinct phospholipids in the autoclaved model emulsions and the commercial formulations is observed (see Tab. 26). The PC:PE ratio of the aqueous phase is enhanced as faster hydrolysis of PE takes place here. As the native model emulsions prepared by Herman [1992] possessed similarly low PC:PE ratios as Intralipid® (and thus higher PE contents) PC redistribution is not observed so strongly for the model emulsion systems and commercial emulsions reported here, containing higher PC:PE ratios than the former. Despite the differences found for the model emulsions, the results for Intralipid® are in agreement with Herman [1992], who however could not verify substantially increased PC:PE ratios for the cream layer of commercial emulsions. The data presented by Groves and Herman [1993] were also only based upon PC and PE concentrations of the separated phases, which underestimates the influence of phospholipid disappearance owing to lyso-phospholipid formation. Furthermore, an interaction or aggregation of vesicular aqueous phospholipids with the dispersed oil phase must also be taken into account (see Chapter 4).
Differences to HPLC data in the literature are probably owing to different sample dilution and detection procedures. Chloroform, which was used by Herman [1992], did not dissolve the amounts of lyso-compounds present in the samples reported here, necessitating use of 2-propanol:n-hexane (1:1 v/v). Herman and Groves [1993] also stated that flame ionisation detection (FID) was difficult to handle, excluding thereby the use of various buffers and solvents. As refractive index detection is not applicable for gradient analysis, evaporative light scattering detection seems most promising to solve analytical problems with phospholipid quantitation, but has also proved to show only narrow response linearity [Balazs et al., 1996]. The availability and the ease of operation of UV detection together with various gradient eluents makes this method still valuable for phospholipid analysis, and the accuracy achieved here is sufficient for the desired description of chemical changes in the emulsions.

HPTLC is frequently used for phospholipid quantitation (as e.g. in Ph.Eur. 1998), but even using flying-spot scanning shows a higher analytical error (~10%) than HPLC analysis, confirming reports by Marmer [1985]. This lets this method seem advantageous only where clear systematic detection errors by HPLC detection inhibited accurate quantitation (e.g. lyso-phospholipid detection). Analysis of various phospholipid compounds in one sample would require considerable effort for additional plates and different staining agents to allow concomitant analysis of standard dilutions and duplicate samples.

Tab. 26 shows the distribution behaviour of PC and PE between aqueous and oil compartments after centrifugation of some model emulsions before and after autoclaving for 20 min at 121°C, as well as for selected commercial emulsions. From the ratios between the phospholipid contents in the cream layer and the aqueous phase, it becomes evident that model emulsions contained more PC and PE associated with the cream layer, which appeared enhanced after autoclaving by a factor of 2-4, compared with commercial emulsions. This can either be explained by formation of larger oil droplets from the smallest ones, thus becoming separated into the cream layer more easily, or by association of additional phospholipids with the oil droplets present, as Groves and Herman [1993] suggested. However, this question could only be clarified by microscopical evidence (see Chapter 4).

Emulsion

PC (mol/g)

PE (mol/g)

Total PC:PE ratio

PC in CL:AS ratio

PE in CL:AS ratio

5%, 1000 bar, 1.2% Lipoid E80®, CL

7.04110-5

6.10910-6

11.5 : 1

9.5 : 1

7.2 : 1

5%, 1000 bar, 1.2% Lipoid E80®, AS

7.44010-6

8.43310-7

8.8 : 1

9.5 : 1

7.2 : 1

5%, autoclaved, CL

6.07510-5

5.98610-6

10.1 : 1

16.4 : 1

17.6 : 1

5%, autoclaved, AS

3.71310-6

3.39610-7

10.9 : 1

16.4 : 1

17.6 : 1

10%, 1000 bar, 1.2% Lipoid E80®, CL

5.31710-5

6.64610-6

8.0 : 1

15.9 : 1

14.6 : 1

10%, 1000 bar, 1.2% Lipoid E80®, AS

3.35210-6

4.55110-7

7.4 : 1

15.9 : 1

14.6 : 1

10%, autoclaved, CL

6.00210-5

7.26610-6

8.3 : 1

42.5 : 1

54.7 : 1

10%, autoclaved, AS

1.41310-6

1.32910-7

10.6 : 1

42.5 : 1

54.7 : 1

Intralipid 10®
(batch 68460-51/CL)

2.67210-5

6.76310-6

3.9 : 1

13.9 : 1

14.7 : 1

Intralipid 10®
(batch 68460-51/AS)

1.92910-6

4.60010-7

4.2 : 1

13.9 : 1

14.7 : 1

Lipofundin 10%N®, CL

3.21410-5

3.20210-6

10.0 : 1

7.6 : 1

7.8 : 1

Lipofundin 10%N®, AS

4.20510-6

4.10610-7

10.2 : 1

7.6 : 1

7.8 : 1

Intralipid 20®
(batch 70178-51),CL

3.00010-5

8.01910-6

3.7 : 1

7.9 : 1

6.7 : 1

Intralipid 20®
(batch 70178-51),AS

3.79010-6

1.19010-6

3.2 : 1

7.9 : 1

6.7 : 1

Intralipid 30®
(batch 88555-71),CL

2.73710-5

8.85310-6

3.1 : 1

6.7 : 1

6.3 : 1

Intralipid 30®
(batch 88555-71),AS

4.11210-6

1.41110-6

2.9 : 1

6.7 : 1

6.3 : 1

30%, 1000 bar, 1.2% Lipoid E80®, CL

2.51910-5

3.98510-6

6.3 : 1

12.5 : 1

15.7 : 1

30%, 1000 bar, 1.2% Lipoid E80®, AS

2.01210-6

2.54210-7

7.9 : 1

12.5 : 1

15.7 : 1

30%, autoclaved, CL

2.69310-5

3.18110-6

8.5 : 1

55.2 : 1

38.5 : 1

30%, autoclaved, AS

4.88310-7

8.26410-8

5.9 : 1

55.2 : 1

38.5 : 1

Tab. 26 PC and PE content of centrifuged oil and aqueous compartment of selected commercial and model emulsions as determined by HPLC analysis (CL-cream layer / AS-aqueous subnatant / values given as mol of lipid per weight of separated phase)

3.3 Conclusions

The film rigidity of egg lecithin films clearly increases with degree of saturation of fatty acids. Although thermoanalytical and monolayer film properties of Lipoid EPC® were similar to the less refined Lipoid E75® and E80®, it showed weaker emulsifying properties. As could be shown by droplet-to-planar-surface coalescence studies, the incorporation of minor components (as contained in Lipoid E75® and E80®) resulted in prolonged droplet stability. The amount of hydrolysis products required for marked reduction of film thinning rates were, however, too high to be solely responsible for the stabilising effect in the emulsions. Zeta potential and particle size measurements implied, however, that the presence of additional sodium oleate resulted in very effective stabilisation of Lipoid EPC® emulsions during homogenisation and autoclaving. A faster adsorption of these more water-soluble amphiphiles to the freshly-created interface could also have accounted for a more effective homogenisation process. High homogenisation pressure for 5-10 cycles yielded finely dispersed emulsions with smaller droplet diameters than commercial samples.
Phospholipids contained in liposomes were shown to hydrolyse faster during heat stress than in association with the oil droplets. Hydrolysis of PE occurred faster than of PC in all cases. It could also be shown that the hydrolysis products produced within emulsions remain largely in association with the oil phase. Hydrolysis products forming during autoclaving and storage should therefore be expected to even enhance droplet stability. pH was found to have an important influence on autoclavability of the emulsions, possibly by maintaining droplet Zeta potential above -50 mV.

Emulsions from Lipoid E80® showed lecithin composition and distribution similar to Lipovenoes® and Lipofundin® formulations. However, Intralipid® emulsions showed a lower PC:PE ratio and contained even more PE than was reported by Kuksis [1985] (Tab. 3). The variability of PC:PE ratios observed did, however, not influence the stability of the emulsions.

Phospholipid relocation from the aqueous to the dispersed oil phase occurred within the emulsions during autoclaving. This points to an interaction of the phospholipids with the oil droplets, with a possible influence on their autoclaving stability [Groves et al., 1985]. Contrary to the reports of Groves and Herman [1993], the relocation of PC and PE did not alter their respective ratios during autoclaving. This contradicts the authors’ suggestion of a cubic phase formation during autoclaving, since these phases are usually observed at higher PE and low water content [Eriksson et al., 1985 / Small, 1986]. The presence of additional phospholipid material at the O/W-interface will be demonstrated in Chapter 4.

Previous PageNext Page


Konvertiert vom Dissertationen Online Team im CCC der Universität Erlangen

dissertationen@ccc.chemie.uni-erlangen.de