Centella cold-process stability: why a better zeta potential can coexist with a changing emulsion

An increasingly negative zeta potential can look reassuring on a formulation report. Yet in an 84-day study at 45°C, a cold-processed water-in-oil emulsion containing a Centella ingredient developed a rising optical instability index and additional coarse-size peaks even as its zeta-potential magnitude increased. The results are not necessarily contradictory: the instruments examined different properties, with very different sample preparation.[1]

The useful lesson from Krzyżostan and colleagues’ 2024 laboratory study is how to distinguish ingredient-induced viscosity changes, optical changes in the undiluted formulation, electrokinetic measurements after dilution, and retention of individual chemical markers. Its particular commercial ingredient and carrier systems do not represent every Centella formulation.[1]

What did “10% Centella” actually mean?

The stability experiments compared an oil-in-water emulsion (O/W), a water-in-oil emulsion (W/O) and a gel containing 10 wt.% CICA EX. The paper identifies this commercial ingredient by an INCI composition of butylene glycol, water, Centella asiatica extract, madecassoside and asiaticoside. The addition level is therefore not 10% pure asiaticoside, nor can it be assumed to mean 10% native dry extract.[1]

The carriers differed substantially. The O/W system used the nonionic Galehemp OW emulsifier, the W/O system used EWO Vegetable Emulsifier, and the gel used a thickener blend containing xanthan gum, sodium stearoyl lactylate, tapioca starch and algin. The authors reported emollient contents of 21 wt.% in O/W and 31 wt.% in W/O. Preparation took place at 25°C, with stirring and homogenization for the emulsions and stirring to form a homogeneous gel.[1]

This was not a single-variable experiment in which only the continuous phase changed. Emulsifier chemistry, lipid composition and structure-building mechanisms changed with the carrier. O/W performed better in this particular comparison; that makes it a useful screening lead, not proof that every Centella ingredient is inherently more stable in an O/W base.

The ingredient changed viscosity before storage began

Table 2 reports initial physicochemical properties, not viscosity losses over 84 days. Adding the commercial ingredient reduced viscosity in all three carriers, while density remained close to 1 g/mL and the authors reported no obvious difference in initial appearance.[1]

Scroll horizontally or focus the table and use arrow keys; row labels remain visible.

Values are means ± standard deviations, n=3, from Table 2. Measurements used spindle SC4-29 for W/O and SC4-21 for O/W and gel, all at 100 rpm. These are apparent viscosities under specified conditions, not complete rheological profiles.[1]

A visually similar sample can therefore have a different flow response. However, the study did not include a fully matched water/butylene-glycol vehicle control without the botanical components. The viscosity decrease cannot be assigned entirely to one triterpenoid. For a new formulation, a more informative comparison would retain the blank base, a matched vehicle control and the complete commercial ingredient, with total formulation mass controlled.

What changed during the 84-day test?

Particle size: O/W was not literally unchanged

Figure 1 shows a bimodal size distribution for O/W. The smaller mode, around 1 µm, was relatively stable. The larger mode moved from 144.5 µm initially to 91.0 µm at day 14 and 76.0 µm at day 84. The authors interpreted the later distributions as relatively stable, but that should not be shortened to “no particle-size change.”[1]

W/O developed small additional peaks around 100 µm and 550 µm at days 56 and 84, indicating the appearance of larger structures. Size measurements alone cannot fully distinguish aggregated droplets from droplets enlarged by coalescence. The authors interpreted these changes alongside the optical measurements as evidence of a greater tendency to destabilize in W/O.[1]

Zeta potential: a heavily diluted sample, not the intact emulsion

The article text reports O/W changing from −36.6 to −59.4 mV, W/O from −33.5 to −43.2 mV, and gel from −21.3 to −28.9 mV. All three had a greater absolute magnitude at day 84 than at preparation.[1]

The measurement method is essential to interpreting those numbers. A 0.05 g portion of emulsion or gel was suspended in 5 g distilled water using an ultrasonic bath. A small aliquot was then diluted again in 15 mL water: 60 µL for emulsions or 120 µL for gel. Electrophoretic measurements were conducted at 25°C and converted to zeta potential.[1]

These results describe electrokinetic behavior after specified dilution and sonication. For an originally oil-continuous W/O emulsion in particular, they are not a direct observation of the bulk structure in its storage container. Zeta potential can contribute to a comparison without serving as a stand-alone stability release criterion.

TSI: a different view of the undiluted formulation

For Turbiscan measurements, samples were placed directly in glass vials. Changes in backscattering and transmission were processed into the Turbiscan Stability Index, or TSI. This preparation was different from the highly diluted zeta-potential measurement.[1]

Figure 3 shows an initial TSI rise in O/W followed by an approximately level trajectory—not an index that stayed at zero or remained unchanged from day 0. W/O continued to rise, reaching 27.7 at day 84. Gel changed more rapidly after roughly a month and reached 15.7 at day 84. The faster change falls between the plotted observations at approximately days 28 and 42—about a month into storage.[1]

The W/O pattern is especially instructive: increasing zeta-potential magnitude, increasing TSI, and late-emerging coarse-size features occurred together. Interpreting each result in light of its sample preparation and measurement target helps determine the next investigation.

Scroll horizontally or focus the table and use arrow keys; row labels remain visible.

Observations are drawn from Figures 1–3, Table 2 and supplementary Figure S1; the development uses in the final column are proposed interpretations.[1][2]

Marker retention is a separate question

The same paper assayed madecassoside and asiaticoside, but on a different timetable. Figure 5 compares freshly prepared samples with samples stored at 45°C for one and two months. Although the authors described the concentration changes as modest, the plotted bars trend downward. These measurements do not establish zero active loss over 84 days.[1]

Physical structure and chemical content can change independently. Aggregation need not cause an equally large loss of the assayed markers, and retained marker content does not make a separating or rheologically altered product acceptable. Keeping these endpoints separate helps determine whether the next change should address the emulsifier, the ingredient vehicle or storage conditions.

There is also an analytical reason not to overinterpret small differences in Figure 5. Table 3 gives a calibration range of 0.03–0.10 mg/mL, while Figure 5 displays extract concentrations around 17–22 µg/mL, equivalent to 0.017–0.022 mg/mL. The methods do not clearly describe a separate lower-range calibration for these samples. The reported asiaticoside spike recovery was 111.95%. These points do not demonstrate that all results are invalid, but they limit confidence in deriving highly precise retention percentages from the graph. They are not a sound basis for calculating a finished product’s shelf life.[1]

Why the release experiments need their own interpretation

The release samples contained either 2 wt.% of each glycoside or 5 wt.% of each: 2% madecassoside plus 2% asiaticoside, or 5% plus 5%. Approximately 1 g of formulation released through a regenerated-cellulose membrane into 100 mL buffer at 37 ± 0.5°C. The acceptor fluids were pH 5.8 or 7.4, and testing lasted 24 hours. These were not simply lower concentrations of the 10 wt.% commercial ingredient used in the stability tests.[1]

Figure 4 reports cumulative percentage release. Several higher-loading formulations released a smaller fraction of their starting content, but that does not mean that adding more released no additional mass. Table 4 provides a useful example: for madecassoside from O/W at pH 5.8 after 480 minutes, the reported fractions were 48.3% at 2% loading and 37.6% at 5% loading. Normalized to 1.00 g formulation, those values correspond to 9.66 mg and 18.80 mg respectively. The released fraction fell, while the mass calculated from the nominal loading increased.[1]

That is a mass calculation using reported fractions and nominal concentrations, not a new experiment. Release through regenerated cellulose is also distinct from absorption through human skin. The practical lesson is to give ingredient addition, individual-marker loading, released fraction and released mass separate columns rather than calling all four “dose.”

Applying the case to cold-process development

The following approach is a development proposal drawn from the measurement differences, not a finished formula validated by the paper.

Separate the immediate addition effect. Compare the blank base, a matched ingredient vehicle and the complete Centella ingredient under consistent final-mass and measurement conditions. Record both the freshly made sample and its state after a defined equilibration period, so early structure formation is not automatically classified as long-term instability.

Locate the change rather than dismissing it. Retain appearance records, microscopy, full size distributions, viscosity conditions and optical profiles. If TSI rises while zeta-potential magnitude remains high, investigate changes with sample height and the formation of coarse structures. A reassuring electrokinetic number should not override those observations. When replacing a liquid ingredient, include its added water, glycols and other constituents in the comparison.

Complete physical, chemical and microbiological evaluations separately. Marker assays need suitable matrix recovery and calibration coverage. Physical testing should address the intended packaging and storage scenarios. Preservation requires its own evaluation. The paper’s 45°C, 84-day physical comparison supplies a screening case, not a universal conversion to ambient shelf life; it does not report a finished-product preservative challenge test.[1]

The most transferable result is not a universal use level. It is an order of interpretation: identify the ingredient and sample preparation, follow each endpoint over time, then decide which part of the formulation needs attention.

For the differences between prepared liquids, refined triterpenoid fractions and isolated compounds, see the Centella ingredient guide. For material selection and sample discussions, visit Centella asiatica extract.

Sources

[1] Controlled Release of Madecassoside and Asiaticoside of Centella asiatica L. Origin from Sustainable Cold-Processed Topical Formulations (2024; full text XML)

[2] Supplementary materials of the same study (original ZIP)

Krzyżostan M, Wawrzyńczak A, Nowak I. Molecules 2024, 29(23), 5583. DOI: 10.3390/molecules29235583