Which test methods are recommended for validating mixing quality in quality control?
A robust validation of mixing quality always combines three elements: representative, documented sampling, product-suitable quantitative analytics, and statistical evaluation. The central metric is usually the coefficient of variation (CoV) or the relative standard deviation (RSD). Methods such as tracer analytics, HPLC, GC, ICP-OES, NIR, Raman, image analysis, particle size analysis or rheological measurements are useful in addition, depending on the product, the critical component and the type of process.
Sampling as the foundation
Even the analytically best method cannot deliver a reliable result if the sampling is not representative. For batch mixers, samples should therefore be taken from different spatial regions and additionally across the entire course of discharge. Critical areas include, for example, wall and base regions, the centre of the mixing chamber, the outlet zone and, with certain designs, possible corners or low-throughflow zones. With continuous mixers, samples are taken staggered over time across a sufficiently long production period, so that dosing and residence-time fluctuations become visible.
Ten individual samples per batch or measurement time point are often regarded as a practical minimum. In regulated or particularly critical applications, significantly more samples may be required. For blend assessment, the FDA recommends stratified sampling from different mixer positions and expects all replicates to be evaluated, so that both differences between positions and the scatter within a position can be statistically assessed. For convective mixers, pharmaceutical recommendations discuss up to 20 spatially distributed sampling points, because corners and the discharge area in particular must be taken into account.
Sample quantity and sampling device must match the particle size, the concentration of the critical component and the later dosing or packaging unit. A sample that is too small can produce high random scatter due to the low number of particles it contains. A sample that is too large, by contrast, can mask local concentration differences. Sampling itself can also segregate the mixture; the probe, sampling speed, sample container, transport and preparation must therefore be qualified. Method variance should be determined separately from the actual process variance.
Statistical evaluation
The most commonly used metric is CoV or RSD:
CoV=RSD=sxˉ×100%
Here, s is the standard deviation of the measured concentrations and xˉ their mean. A low value indicates low relative scatter of the target component under investigation. In many technical mixing applications, values below about 5 per cent serve as an indication of very good homogeneity, while values up to approximately 10 per cent can often still be acceptable. For food and feed mixtures, work is often done with ten samples and a CoV below 10 per cent for a suitable marker. These values, however, are not a universal requirement: for vitamins, allergens, highly concentrated active ingredients, dyes or other critical minor components, stricter, product-specific criteria can be necessary.
In addition to the CoV, the mean against the target value, individual values, minimum and maximum, range, confidence intervals, and the spatial and temporal distribution should be taken into account. A low CoV, for example, can occur despite systematic under-dosing, if all samples are evenly but too low in concentration. Likewise, individual critical outliers can be masked by a good overall value. Mixing indices such as the Lacey index can theoretically describe the mixing state between a segregated starting condition and a random mixture, but in industrial quality control they are usually a supplement, not a replacement, for product-specific specifications and CoV evaluation.
Analytical methods
Tracer methods are a good starting point for many solid mixtures. An analytically well-detectable marker component is used, for example salt or chloride, a mineral, a dye, magnetic particles or another formulation-compatible tracer. The concentration is then determined in the individual samples. The marker must represent the mixing behaviour of the most critical component as closely as possible. A coarse-grained, free-flowing salt marker, for example, is only of limited suitability if the critical real component is a fine, cohesive powder.
Chemical-analytical methods are suitable for the direct determination of target components. HPLC or GC are used where organic active ingredients, flavours or additives need to be determined selectively and quantitatively. ICP-OES, ICP-MS or AAS are suitable for elements and mineral aggregates. Titrations can be used for reactive or ionic components. For moisture, Karl Fischer titration, halogen moisture determination or gravimetric drying methods are common. These methods are generally highly selective and precise, but are often carried out off-line and place high demands on sampling, sample preparation and laboratory organisation.
NIR spectroscopy is a suitable method for many powders, granules, moist mixtures and pastes for fast at-line, online or inline monitoring. With suitable calibration, it can capture the concentrations of critical components, moisture and the time course of homogenisation. NIR is regarded as a powerful tool for real-time assessment of mixing homogeneity and can also detect inhomogeneous batches. Raman spectroscopy is likewise suitable where the components have sufficient Raman signatures. Both methods require a chemometric model validated against reference analytics, for example based on partial least squares regression or principal component analysis. They are particularly suitable for defining a mixing endpoint and detecting batch deviations, but do not straightforwardly replace qualified reference sampling.
Image-analysis methods are suitable for mixtures with optically distinguishable components. Camera systems can capture colour distribution, visible agglomerates, particle count, surface coverage or layering. They are helpful with spice blends, cereals, granules, powders containing dyes, and pastes and coatings. Their informative value is limited to the visible surface or a defined optical measurement zone; for a complete volumetric assessment, sampling and image evaluation must be planned accordingly.
Particle size analyses by sieving, laser diffraction or dynamic image analysis are particularly important where segregation by size, shape or breakage fraction is suspected. They do not directly show chemical homogeneity, but can demonstrate whether the particle structure has changed through mixing, or whether different fractions appear at different times in the discharge. With liquid, pasty and multiphase systems, viscosity, yield stress, conductivity, density, turbidity or pH measurements supplement the concentration analysis. They are useful as indirect homogeneity indicators, but do not necessarily prove the distribution of a specific critical component on their own.
Validation strategy
A robust validation begins with a mixing kinetics study. The same formulation is examined at several mixing times, for example after short, medium and long mixing times. For each time, spatially and temporally distributed samples are taken, analysed and statistically evaluated. This allows the determination of when the target homogeneity is first reached and whether segregation, particle damage or unwanted heating occurs with longer mixing. The result should be a process window, not just a single target value for the mixing time.
In the next step, robustness against realistic fluctuations is checked: batch size, fill level, raw material batch, moisture, particle size distribution, dosing sequence, rotational speed, liquid addition and discharge conditions. A design-of-experiments approach can help identify the critical process parameters and their interactions. Checking the discharge is particularly important. A mixture can be homogeneous in the mixer but segregate again during emptying, along drop paths, in conveying systems or in packaging. At least samples at the start, middle and end of the discharge should therefore be examined.
Testing, monitoring and documenting mixing quality
Based on numerous documented mixing quality tests with very good results, amixon® mixers are regarded as fundamentally validated in mixing performance. The proven mixing principles, in particular the three-dimensional SinConvex® forced restratification, are designed to produce a technically ideal random mixture. For many standard tasks, extensive preliminary trials on basic mixability can therefore be dispensed with. Additional mixing trials are nevertheless useful and, in certain cases, necessary: for example with new or particularly critical formulations, at very low dosing fractions, with strong differences in particle size or bulk density, with cohesive or moisture-sensitive powders, and where the minimum possible mixing time needs to be determined with high confidence. The goal is not to re-prove the apparatus's basic mixing capability, but to establish the shortest robust process window for the specific formulation.
A robust mixing quality assessment begins with representative sampling. Metrics such as the coefficient of variation (CoV) and the relative standard deviation (RSD) are only meaningful if the samples represent the entire batch. CoV and RSD are calculated identically: CoV=RSD=(s/xˉ)×100%, where ss is the standard deviation and xˉxˉ is the mean of the measured target component. Low values indicate low scatter and therefore an even distribution of the component under investigation. The permissible values are defined on a product-specific basis. As a technical guide, CoV or RSD values up to about 5 per cent often stand for very good homogeneity; values up to approximately 10 per cent can still be acceptable in many applications. However, the formulation, analytical method, dosing fraction, number of particles per sample and the respective quality or customer specification remain decisive.
In the amixon® pilot plant, the sampling plan is developed together with the operator. Sampling is carried out in a stratified manner, that is, across different positions in the mixing chamber and over the time course of discharge. Discharge in particular is decisive: a mixture can be homogeneous in the mixer and then segregate again along drop paths, in conveying lines, through vibration, or during filling. Samples should therefore be examined not only in the mixer, but also at the start, middle and end of discharge. Typically, at least ten individual samples per batch or measurement time point are used. A larger number of samples may be required for complex or regulated applications. Stratified sampling from multiple areas and process phases is an established principle for investigating homogeneity.
On batch mixers, low-dead-space discharge elements and the good accessibility provided by large CleverCut® inspection doors support sampling at defined locations and times. Besides the spatial distribution of samples, sample quantity, sampling device, particle size, dosing fraction and analytics must be matched to one another. A sample that is too small can produce high random scatter; a sample that is too large can mask local inhomogeneities. The variance of sampling and laboratory analytics must also be assessed separately from the actual process variance, so that analytical uncertainty is not wrongly interpreted as a mixing fault.
At the start of long process routes, the amixon® Gyraton® silo mixer GM can be particularly helpful. Large batches are homogenised early and reproducibly, before undergoing further conveying, storage, dosing, conditioning or filling steps. This creates a defined, stable starting basis for downstream processes. As a result, the effort for accompanying individual analyses and permanent fine adjustments further along the process chain can be significantly reduced, because fluctuations do not first need to be detected and compensated for at later production stages. The Gyraton® GM is suitable for large batches of approximately 10 to 100 m³ and can be operated at low mechanical stress. The actual suitability, mixing time and homogeneity are established through trials for the product, batch size and downstream process route.
The choice of analytics depends on the most critical formulation component. Tracer methods are a pragmatic solution for many powder mixtures. Salt, a mineral, a dye or another selectively detectable, formulation-compatible component can serve as a marker, for example. HPLC or GC can be used for organic active ingredients, flavours or additives. ICP-OES, ICP-MS or AAS are suitable for mineral and metallic constituents. Moisture distributions can be assessed, depending on the application, with Karl Fischer titration, halogen moisture determination or gravimetric drying methods. For optically distinguishable mixtures, image analysis can provide additional information on colour pattern, layering, particle distribution or agglomerates.
NIR or Raman spectroscopy can be used for direct process monitoring. NIR is suitable for many powders, granules, moist mixtures and pastes where the critical components or the moisture can be captured spectrally. Raman can be advantageous where individual components have characteristic Raman signals. Both methods can track mixing progress in real time and help identify the mixing endpoint. This requires a chemometric model calibrated and validated against representative reference analytics. NIR is used as a powerful method for real-time assessment of mixing homogeneity.
For process control, adherence to the validated parameters is particularly important. Mixing time, rotational speed or circumferential speed, fill level, raw material sequence, dosing quantities, dosing times, liquid addition and, where applicable, temperature profiles are stored in the PLC. The most effective ongoing monitoring of a once-established formulation is control of these critical process parameters. Deviations from the target time, target speed or other limit values can be detected, alarmed and documented immediately. PLC monitoring, however, does not replace analytical verification for new, changed or particularly critical formulations.
Mixing programs can be integrated into the operator's ERP or manufacturing execution system. Barcode or RFID capture allows formulation, raw material batch, container identity, mixing parameters, operator interventions, cleaning status and batch release to be linked traceably. This supports traceability, deviation management and validation. In GMP-oriented projects, amixon® can support DQ, IQ and OQ; documentation and automation can be aligned project-specifically with EU-GMP and FDA 21 CFR Part 11.
Warning signals must always be investigated. These include fluctuating analysis values, noticeable differences between discharge samples, visible segregation, uneven moisture, batch deviations, unexpected torque or power consumption, changed bulk density, or instabilities in downstream dosing and filling processes. Causes can include an unsuitable mixing time, fluctuating raw materials, a dosing error, segregation after the mixer, an unsuitable fill level, or non-representative sampling. Trials with the original product help to specifically distinguish these causes and adjust the process window.
The product-gentle design of amixon® mixers supports reproducibility. The tool circumferential speed can be set within a range of approximately 0.8 to 3.5 m/s, depending on the design and mixing task. SinConvex® forced restratification guides the product upward at the edge and back downward under gravity at the centre. The mixing action is thus generated through controlled circulation rather than pronounced throwing, impact or crushing zones. Sensitive particles, coatings and agglomerates can thereby be preserved with limited heat input. Where targeted de-agglomeration is required, cutting rotors can be switched on locally and only during a defined process phase.
Mixing chambers welded free of crevices and ground smooth, mixing tools supported only at the top without a lower shaft passage in the product area, large CleverCut® inspection doors with OmgaSeal® seals, low-dead-space outlet fittings and integrated washing lances support cleaning, accessibility and product changes. Dry and wet cleaning can thus be designed and validated on a product-specific basis. Depending on the project, a design to EHEDG as well as taking FDA hygiene guidelines and 3-A Sanitary Standards into account is possible.