The Single-Use Mixing Study
Researchers at ten23 health tested three mixers in the same bottles, on the same drive: a conventional magnetic stirrer, a bottom-mounted single-use mixer, and a top-mounted one with a fixed baffle. They measured how long each took to dissolve a salt load. They measured how much damage each did to a model protein over four hours. The paper runs fourteen pages in the Journal of Pharmaceutical Sciences.
ten23 health AG, Basel, Switzerland. Andrea Allmendinger is additionally affiliated with the Institute of Pharmaceutical Sciences, University of Freiburg.
Ali D, Plaga JF, Adler M, Mahler HC, Allmendinger A. J Pharm Sci. 2026;115:104448. Published by Elsevier Inc. on behalf of the American Pharmacists Association. DOI: 10.1016/j.xphs.2026.104448
This research was funded by SaniSure. Two of the three mixing systems evaluated are SaniSure products. The experimental work was designed and conducted by the authors at ten23 health in Basel; the flow simulations were executed with support from SaniSure personnel, who are named in the acknowledgements.
One author disclosed an editorial role at the journal and was not involved in the editorial review or the decision to publish. The full funding and competing-interest statements are in the report.
Chosen on Availability, Justified Later
Mixing in single-use bottles runs through much of a fill and finish workflow: thawing and pooling of drug substance, dilution, and formulation ahead of filtration and filling. It is routine enough that the device doing it is usually selected from whatever is already in the suite. The justification gets written afterwards, at qualification, out of whatever data can be assembled at the time.
This paper is that data. It compares three systems on the same PharmaTainer bottles and the same drive mechanism: a conventional magnetic stirrer, a bottom-mounted four-blade single-use mixer, and a top-mounted bar impeller with an integrated fixed baffle. Tip speed is matched rather than rotational speed, so impellers of different diameters are compared on equal terms.
What comes out of it is a characterisation rather than a recommendation: two statistical models, one for how quickly each system reaches homogeneity and one for what each does to a model protein, and a flow simulation that explains why the two models disagree about which variables matter.
The four numbers this study is usually quoted for, and what they do and do not establish, are summarised on the single-use mixing and protein stability insight page. This page is where you request the paper itself.
Find the Decision You Are Making
- Selecting a mixer for a new process and wanting a basis for the choice that is not simply what the suite already stocks.
- Justifying an incumbent magnetic stirrer to a reviewer who has asked why it was chosen and what was compared against it.
- Running bottles at partial fill during early development, small-batch work, or the tail of a pooling step. Fill volume is one of only two variables that moved product quality in this study.
- Building a qualification package and needing to know which conditions published data actually covers before citing it.
- Scaling a mixing step out of development and into 5 L and 20 L formats, where the three systems behaved differently from one another.
Six Questions the Paper Answers About Your Process
- How much faster is a purpose-designed mixer at matched tip speed, and how wide is the spread? Reported as an average with its full range across configurations, not as a single headline figure.
- Which variables actually move dissolution time? Table 3 gives estimates, standard errors, p values and significance for mixer type, impeller-to-bottle diameter ratio, tip speed and fill volume ratio.
- Which variables move protein quality? Table 4 does the same for the protein inactivation rate. It is not the same list, and the difference between the two lists is the most useful thing in the paper.
- Does running faster cost product? The model answer and the mixer-specific exception are both stated, and they point in different directions.
- What happens at low fill volume, and what happens at high? The gap between mixers widens at one end of the range and closes completely at the other.
- Why does any of it happen? Velocity vector fields and shear stress contours for all three systems, plus the correlation between shear stress per unit volume and measured inactivation.
What Is Actually in the Fourteen Pages
Read the two regression models against each other and one thing stands out. Dissolution time was governed by mixer type, tip speed and impeller-to-bottle ratio. Protein quality was governed by mixer type and fill volume. Mixer type is the only predictor that reached significance in both. Everything else moves one outcome and leaves the other alone.
Our reading is that this makes mixer selection a process parameter rather than a consumable choice, and one worth characterising during development rather than settling at procurement.
That observation is ours, not the authors’. They report each model in its own table and leave the comparison to the reader. Both tables are in the report, so you can check it against the source rather than taking it from us.
Where the Study Does Not Reach
Every study has boundaries, and the useful ones say where they are. These are drawn from the authors’ own statements. If you are using this work to support a qualification decision, this is the part that matters most, and it is a section of the report rather than a footnote to it.
- At the highest fill volume tested, the mixers were indistinguishable. No relevant differences between the three systems, nor against an unagitated control. Mixer selection mattered at low fill volumes. At high fill, on this measure, it did not.
- The three mixers were not sampled equally. The bottom-mounted mixer was tested at three fill volumes, the other two at minimum and maximum only. The authors state that this limited the model’s ability to estimate the effect of the magnetic stirrer and the top-mounted mixer.
- The low-fill comparison is not exactly like for like. At minimal fill the bottom-mounted mixer head was not fully submerged, while the other two systems stayed submerged. Low fill is precisely where the largest differences appeared.
- Shear alone does not explain protein damage. The correlation between calculated shear stress and measured inactivation returned an R-squared of 0.72. The authors are explicit that interfacial phenomena, cavitation, local thermal effects and contact between moving parts all contribute.
- Temperature was assumed to be uniform. Local temperature fields were not measured for any mixer, and the authors say this assumption needs to be considered when interpreting the relative effects of different configurations.
- The fluid was water and the protein was a surrogate. Lysozyme at 0.1 mg/mL held at 72 °C, and aqueous fluid properties in the simulation rather than real formulation viscosity. Not a therapeutic protein under manufacturing conditions.
- The study was funded by SaniSure. It was designed and executed by ten23 health and passed peer review at the Journal of Pharmaceutical Sciences. We would rather you read it and judge it than take our summary for it.
Written for process development, MSAT, and fill and finish teams selecting or defending a mixing system, and for qualification and validation groups who need to know which conditions a published comparison actually covers before citing it in a package.
Requesting the Single-Use Mixing Study
What exactly do I receive?
The complete 14-page paper as a PDF, exactly as published in the Journal of Pharmaceutical Sciences. It is delivered on the next page immediately after the form, and a copy is also sent by email. It includes both regression tables, the dissolution and protein data, the flow simulation figures, the authors’ limitations, and the funding and competing-interest statements.
Which mixing systems does the study compare?
Three. A conventional magnetic stirrer, a bottom-mounted four-blade single-use mixer operating between 200 and 600 rpm, and a top-mounted bar impeller with an integrated fixed baffle operating between 80 and 400 rpm. All three were magnetically coupled with suspended mixing devices, run in the same PharmaTainer bottles on the same stirrer plate, and compared at matched tip speed rather than matched rotational speed.
What scale does the single-use mixing study cover?
Four PharmaTainer bottle sizes: 0.5, 2, 5 and 20 liters, at fill volumes from 0.1 to 20 liters. The magnetic stirrer and the bottom-mounted mixer were tested across all four sizes. The top-mounted mixer is only available in larger formats, so it appears in the 5 L and 20 L configurations only.
Does the study cover my formulation?
Almost certainly not, and the report says so. The experimental protein was lysozyme at 0.1 mg/mL in phosphate buffer with polysorbate 80, held at 72 °C, and the flow simulations used aqueous fluid properties rather than real formulation viscosity. Lysozyme is a surrogate chosen because it is known to be sensitive to mixing at that temperature. The study characterises mixing behaviour; it does not predict the stability of a specific therapeutic protein.
Who funded the single-use mixing study, and how do I verify the disclosures?
SaniSure funded the study, and two of the three systems tested are SaniSure products. The work was designed and executed by researchers at ten23 health in Basel, none of whom are SaniSure employees, and it passed peer review at the Journal of Pharmaceutical Sciences. The funding statement, the competing-interest declarations and the acknowledgements naming SaniSure’s role in the simulations are printed in the report itself, so every disclosure summarised on this page can be checked against the source document.
Who is this report for?
Process development, MSAT, and fill and finish teams selecting or defending a mixing system, and qualification and validation groups who need to establish which conditions a published comparison actually covers before citing it. It assumes familiarity with mixing terminology such as tip speed and impeller-to-bottle diameter ratio, and it reports statistics rather than summarising them.