FRR - Final Report

Spray drying is a widely used and well-established process across many industries, including food, pharmaceuticals, cosmetics, and chemicals. The process offers numerous benefits, such as reducing microbial growth, minimizing enzymatic degradation reactions, and significantly reducing the final volume of the product. This makes spray drying an important tool for production of high-quality, stable powders suitable for storage, application, and transportation (Dantas et al., 2023).

In the spray drying process, a solution, suspension, or emulsion is atomized into fine droplets, which are then exposed to a heated carrier, such as air or superheated steam, in the drying chamber (Dantas et al., 2023; Sobulska et al., 2022; Walton & Mumford, 1999). As a result of this interaction, the droplets rapidly lose moisture, which leads to the conversion of the liquid into solid. The fundamental criterion of spray drying processes classification is the method by which the atomized material comes into contact with the drying medium. This can take form of a co-current, counter-current, or mixed flow mode. Drying systems that operate in co-current flow are often preferred in the industry due to simple construction setup and easy control of the process. Unlike counter-current dryers, co-current systems enhance safety for thermosensitive materials by preventing contact of the dry material with hot inlet air (Zbicinski & Piatkowski, 2009a).

Most spray-dried products can be divided into three main categories, according to the morphology of their particles: skin-forming materials, porous materials, and materials with crystalline structure. Porous materials, also known as agglomerates, form particles bound by submicron dust or a binder, typically with a regular, highly spherical shape and with minimal surface irregularities. The drying process involves gradual solvent evaporation from within the particles, facilitated by the highly porous structure. This prevents significant pressure build-up inside the particle and thus avoids deformation or expansion, so blowholes and cratering are not commonly observed features. Porous materials include silica, colloidal carbon, cocoa, and detergents (Zbiciński & Kwapińska, 2003).

Materials with crystalline structure exhibit highly ordered arrangements of atoms or molecules, forming solid structures composed of large individual crystal nuclei bound together in a microcrystalline phase. The morphology, shape, and size of particles depend on the type of substance and drying conditions. Both solid and hollow particles occur. In some cases, significant internal pressure can develop from the evaporation of solvents within a particle. This can lead to disruptions, resulting in the formation of craters and secondary nucleation centers. Examples of materials with crystalline structures include sodium chloride, sodium carbonate, zinc sulfate, sodium pyrophosphate, sodium benzoate, and sodium formate (Walton & Mumford, 1999).

In skin-forming materials, drying initially occurs on the droplet's surface, increasing local viscosity and leading to the formation of a thin, hard outer layer known as a 'skin' or 'shell.' This layer is composed of a continuous, non-liquid phase, either polymeric or sub-microcrystalline. These materials typically form spherical particles with smooth surfaces and may be either hollow or solid. Hollow particles, are susceptible to collapse after drying, unlike solid particles, which retain their structural integrity. Surface-active molecules facilitate skin development by accumulating at the phase interface. An increase in temperature, which leads to a higher evaporation rate, also accelerates this process. Rapid skin formation can cause the trapped gas within the particle to expand and potentially rupture, leading to disintegration of the particle. Additionally, residual moisture in the particles may lead to the formation of secondary bubbles. The drying kinetics can lead also to other phenomena, such as particle inflation, shrinkage, crater formation, agglomeration, cracks and gaps, as well as particle vacuolation. This stands in stark contrast to the relatively narrow spectrum of morphological features typically observed in agglomerate and crystalline structures, which generally exhibit cracks, occasional crater formation, blowholes, and hollow particles. According to Walton et al. (Walton & Mumford, 1999), the ability to modify structural morphology of skin-forming particles during the drying process is instrumental in achieving extensive diversity and applicability, mainly in food industry. Skin-forming materials include: sodium silicate, sodium dodecyl sulfate (SDS), potassium nitrate, gelatin, skim milk, chicken eggs and maltodextrin (Walton & Mumford, 1999; Zbiciński & Kwapińska, 2003).

Maltodextrin, a water-soluble carbohydrate, plays a prominent role in the food and pharmaceutical industries, offering diverse functionalities, serving as a carrier for flavors and active agents (Sultana et al., 2018), functioning as an emulsifier (Bae & Lee, 2008; Rowe et al., 2006), filler, or substitute for lactose (Hofman et al., 2016). Thanks to its morphological properties, it offers a solution to the degradation challenges frequently faced by powdered products susceptible to caking or stickiness, primarily caused by the presence of low-molecular weight sugars with a low glass transition temperature (Koç & Kaymak‐Ertekin, 2014). Maltodextrin is used as a surface material in microencapsulation of sensitive components, including β-carotene (Loksuwan, 2007), avocado oil (Bae & Lee, 2008) and nutraceutical extracts (Sansone et al., 2011), and serves as an additive to increase the glass transition temperature of products such as honey (Samborska et al., 2015), sumac extract (Caliskan & Nur Dirim, 2013), or strawberry juice (Gong et al., 2018).

An inappropriately carried out drying process can deteriorate the final product’s quality, which is why precise process control is critical for obtaining high quality product. The nutritional and physical properties of food powders, comprising those containing maltodextrin, include aspects such as taste, aroma, color, and particle: agglomeration, density, porosity, dissolution rate, surface properties and size (Anandharamakrishnan & Ishwarya, 2015; Dantas et al., 2023; Hofman et al., 2016; Koç & Kaymak‐Ertekin, 2014). These properties can be altered by manipulating drying parameters, including temperature and flow rate of the drying medium, atomization method, and overall apparatus design, as shown by numerous published studies (Anandharamakrishnan & Ishwarya, 2015; Caliskan & Nur Dirim, 2013; Dantas et al., 2023; Koç & Kaymak‐Ertekin, 2014; Sobulska et al., 2022; Walton & Mumford, 1999; Zbicinski & Piatkowski, 2009a). In addition, it has been demonstrated (Takeiti et al., 2010) that a significant determinant influencing the particle properties of maltodextrin is the source of the starch hydrolyzed to obtain maltodextrin, along with its dextrose equivalent (DE) value. Recent works on mixtures of maltodextrin with other ingredients, such as proteins or oils, has emphasized the role of maltodextrin concentration in droplet dispersion, which is one of the factors that determine the specification of the resulting powders (Bae & Lee, 2008; Both et al., 2020). Nevertheless, the influence of these parameters is still complex and not sufficiently understood. This research is promising not only for optimizing powder quality but also for improving the energy efficiency of spray drying and encapsulation as well as for exploring potential applications.

Research work performed on a spray drying tower constructed at the Technical University of Lodz has been conducted for an extended period. A substantial body of literature exists that describes the impact of process parameters on the characteristics of the resulting powders in co- and counter-current systems (Zbiciński & Piątkowski, 2004) (Zbicinski et al., 2002), (Kwapińska & Zbiciński, 2005). Additionally, publications described experiments conducted using spray drying, which investigated the quality of products obtained by modifying the process itself. These modifications included the introduction of new parameters, such as foaming the sprayed solution (Rabaeva & Zbiciński, 2010),(Lewandowski et al., 2019), introducing a swirl of drying air (Wawrzyniak et al., 2020), using two levels of spray nozzles (Wawrzyniak et al., 2024), flame drying (Piatkowski et al., 2014) or conducting a microencapsulation process of oily substances (Lewandowski et al., 2020), (Adamiec & Marciniak, 2006). However, these studies did not fully analyse the level of significance of individual process parameters on the quality of the obtained product, nor did they focus on determining the significance of the effect of the state of the atomised solution on powder formation. What effect does a sudden change in the physical properties of the atomised solution, occurring, for example, as a result of an accident, have on the quality of the obtained product and the course of the drying process?

The research program presented for IFPRI assumes finding the relationship between the rheological properties of the solution and the drying speed on the morphology of the particles obtained by the spray drying method. Therefore, the following tasks were set during the project:

• Selection of suitable experimental media and determination of quality criteria.
• Measurements of rheological properties of aqueous solutions of selected materials.
• Adaptation of the existing equipment to the project requirements.
• Design of particle-free fall SDD measurement system.
• Carrying out spray drying experiments on the semi-industrial scale.
• Analysis of the physicochemical properties of the obtained powder samples.
• Preparation of a monodisperse droplet generator to construct devices to measure drying kinetics.
• Analysis of the influence of the rheological properties and conditions of the spray drying process on the morphology of the melts obtained in the experiments.
• Experiments to determine the temperature of particles during free fall using the IRTUC (InfraRed Temperature for Unknown Coefficients) method.

This 4 year project (3 years plus one extra year due to Corona) aims in developing a system engineering approach for understanding, optimizing and scaling industrial dry grinding processes, with a special focus on the manipulation of the material properties and, thus, the grinding and classification efficiency by the use of grinding or flow aid substances. In terms of process units for its first phase, the project deals with dry grinding circuits consisting of a ball mill and air classifier. The experimental effort done so far was conducted on a pilot scale circuit.

From the literature and from experiments conducted, it was noted that during milling operations, grinding aids (GA) affect the process mainly in:

• tendency of fine particle agglomeration;
• amount of material coated on equipment surfaces;
• powder flowability;
• total mass of product inside the mill and mean residence time;
• as well as product fineness after grinding.

In terms of the air classification process, it was observed that grinding aid show little to no direct impact of particle cut size. However, they play a stronger role on improving classification efficiency, by improving particle dispersion on air streams and promoting lower recirculation of smaller then target size particles back into the milling stage.

The project work was divided in four main work packages:

1. Identify which aspects of the ball milling process are affected by different grinding aids
2. Identify which aspects of the air classification process are affected by different grinding aids
3. Validate the previous findings on closed circuit milling operations and observe the grinding aid impacts on individual units combines in the overall closed circuit process
4. Modify process models from the literature to account for the presence of GA in the systems behavior, implement a flowsheet simulation tool and validate the flowsheet with mill-classifier circuit data

Despite the widespread use of screw feeders in industry for transporting, metering, and processing powders, scientific understanding of their functioning is far from satisfactory. This is primarily due to the complexity of the mechanical response of powders and the complex geometry of screw feeders. In this project we have developed two models for the kinematics and mechanics of non-cohesive powders.

The first model

relies on several simplifying approximations, but makes the interesting prediction that the feed rate is maximum for a specific value pitch to diameter ratio of the screw.

The second model

is a more detailed continuum model that predicts the variations of velocity, packing fraction and stress within the feeder.

To validate the model predictions, we have conducted DEM simulations and gathered experimental data using a custom-built screw feeder apparatus. Overall, we find good agreement between the experimental data, DEM simulations, and model predictions for non-cohesive powders (such as glass beads). In particular, the simulations and experiments verify the model predictions of the feed rate being maximum at a certain ratio of pitch to diameter of the screw. The experiments and simulations also throw light on the variability of the feed rate, an aspect that is of concern to industry.

We have conducted some experiments and DEM simulations for cohesive powders commonly used in the pharmaceutical industry. We find that the main source of variability in the feed rate is not within the feeder but at the inlet to the feeder. More detailed investigation of cohesive powders, including extension of the model to account for cohesion, remains to be done, and are the primary goals of our endeavours in the next 3 years of the project.

The project brief included the following objectives:

• Identification of appropriate test powders and characterization of their relevant physical and chemical properties.
• Establishment of a test method to quantify material adhesion on compaction tooling over an industrially relevant range of process and environmental conditions.
• Identification of key factors affecting the amount and/or rate of powder adhesion on
• compaction tooling such as: molecular, crystal, surface, and mechanical properties of the powder, surface finish and chemistry (including coated surfaces), process conditions (e.g., pressure/stress) and environmental conditions (temperature, relative humidity)
• Establish predictive criteria for the propensity of adhesion given a set of molecular/crystal properties and process/environmental conditions.

The sticking behaviour was characterized for the following materials: Ibuprofen, Acetylsalicylic Acid (Aspirin), Acetaminophen (Paracetamol), Mannitol (Pearlitol), Sorbitol (Neosorb), Maize Starch B and Microcrystalline cellulose (Microcel) which was used as a reference non-sticking material.

In addition to standard compaction studies, two new test methods were developed: heated die and high-rate compaction. The experimental conditions for a diagnostic test were established.

The key factors controlling sticking were determined as: compaction pressure, compaction rate, temperature and relative humidity.

The team at Leicester conceived a predictive method that uses machine learning to determine the sticking probability of sticking of any existing or new chemical entity from molecular information as follows. The chemical formula together with the structure of the molecule is encoded in SMILES (Simplified Molecular Input Line Entry System), for which molecular descriptors are calculated (Mordred). Machine learning tools including linear discriminant analysis (to rank the descriptors), feature engineering (to balance the data set), principal component analysis (to determine weighting for the descriptors), and support vector machine (to classify sticking and assign probability).

The experimental data generated in the project was used to train the algorithm, together with known sticking and non-sticking materials from the literature. The sticking probability was determined for the materials published in the Handbook of Pharmaceutical Excipients and molecules proposed by the industrial partners.

As a community, our ability to understand and computationally predict wet granulation process has rapidly improved over the past decades. Despite these advancements, we are yet to use models for granulation or other particulate processes to their full advantage, to enable the predictive design of granular product performance.

The aim of this research is to address this challenge to link process and product performance models for wet granulation. To achieve this aim, a unique granule performance model has been developed. Critically, this multi-scale model for swelling driven granule disintegration and dispersion has been developed with the mission to be suitable for linking with existing process models for wet granulation.

A single granule model for the swelling of a granule containing disintegrant has been proposed, with two variations: mono-sized constituent particles, and distributed constituent particles. This model has then been coupled with a population balance model, to allow the evolving particle size of a dispersing granule population to be modelled.

These models have been validated using novel experiments developed in collaboration with the University of Strathclyde, and the results of these experiments have not only provided the opportunity to parameterise these models but have also provided further insights into the rate processes of granule disintegration and dispersion. By varying between formulations of microcrystalline cellulose and dibasic calcium phosphate, the relative contributions of erosion and swelling driven dispersion are elucidated and qualified mathematically. Model assumptions have also been tested through this experimental validation. A key assumption in the population balance model is the dispersion of granules directly into primary particles, with a lack of intermediate granules or aggregates. On testing, it was found that the assumption was valid for dibasic calcium phosphate, but not as suitable for microcrystalline cellulose.

A comprehensive global sensitivity analysis of the models has been performed, and model parameters have been ranked for their importance to the model. This enables the user to prioritise the accurate measurement or estimation of key parameters, such as granule porosity and disintegrant diffusivity, and deprioritise lower ranked parameters.

The development of this model is expected to enable the linking of process and product models for wet granulation in the second phase of this this IFPRI project, and the subsequent use of inverse methods to enable true model driven process design for desired granule performance.

The report covers work from 2019 to 2023, initially focusing on broadening rheological responses in colloidal suspensions by altering the properties of their components. This aims to incorporate a wide range of behaviors into simple formulations, essentially achieving more with simpler mixtures. It highlights the significant impact of particle roughness on the shear thickening of stable suspensions and the more pronounced effects in colloidal gels. Rough particle gels demonstrate unique characteristics such as higher strain or stress thresholds for yielding, flow-independent porosity, and rapid recovery without thixotropy, attributed to hindered rolling motion of particles and clusters. Similar outcomes are achieved with anisotropic particles like rods, beneficial for preventing sedimentation.

The report then details advanced methods introduced for analyzing microscale rearrangements and correlating them to local forces. These include ultra-high-speed confocal rheometry, high-frequency rheology, orthogonal superposition rheometry, and AFMbased techniques. The application of these methods is demonstrated in simplified industrial systems, such as concentrated hydrogel dispersions (Carbopol) and paracetamol dispersions, offering valuable insights.

Lastly, it introduces a minimalistic model for pasty materials, combining a viscoelastic Maxwell model with stress-dependent relaxation time. This effectively describes the transition in pasty materials from elastic to plastic deformation, eliminating the need for a yield stress concept.

In the field of granular rheology, one of the most promising advances of the past decade has been the development of various nonlocal rheologies [1–7]. These constitutive models hold the promise of permitting the determination of a small number of empirical parameters for a particular set of particles, which then can be used to predict flow fields and stresses over a large range of intermittent, creeping, quasi-static, and intermediate flows. In order for these models to be useful, the aim is to make a set of flow measurements for a set of particles in one geometry, and then determine the constitutive parameters for use in predicting flows in other geometries (for the same particles). Doing this requires a quantitative understanding of which properties are set by both the particle properties, and the boundary conditions at the walls.

In Years 1-3, we established that NLR successfully models granular flows across different packing densities, particle sizes and shapes, and shear rates, using just 3 constitutive properties (A, b, μs), but that we must know the amount of slip at the wall from geometry-dependent measurements. During Years 4-7, we extended these measurements to compare which flow properties are set by the particle properties, versus by the wall properties. We performed experiments in both the original annular rheometer, as well as in a vertical hopper, using six different boundary conditions. We found that the roughness and compliance of the boundary strongly controls the amount of wall slip. Nonetheless, we find that we can successfully capture the full flow profile using a single set of empirically determined model parameters, with only the wall slip velocity set by direct observation. Through the use of photoelastic particles, we observed how the internal stresses fluctuate more for rougher boundaries, corresponding to lower wall slip, and connected this observation to the propagation of nonlocal effects originating at the wall. Our measurements indicate a universal relationship between dimensionless fluidity and velocity. The measurements in the annular rheometer are echoed by less-quantitative measurements performed in the vertical hopper. These results have been published in one paper [22] and one preprint [25]. Two graduate students received their PhDs, and one undergraduate gained research experience.

Three IFPRI collaborations have been supported during this seven year period, with Nathalie Vriend, Karen Hapgood, Prahbu Nott and their research groups. With Vriend, we extended our efforts into a chute flow which provided us data at higher inertial number than was possible in the annular rheometer. We defined a quantitative measure for the rate of generation of new force chains and found that fluctuations extend below the boundary between dense flow and quasi-static layers, as well as evaluating several existing definitions for granular fluidity [8]. With Hapgood, we performed stress visualization within 3d printed particles with realistic shapes, using photoelasticity. We characterized the importance of controlling the relative orientation of the print layers and the loading force, and observed that semi-quantitative measurement of internal stresses is possible, with some caveats [9]. With Nott, we performed laboratory tests of an additional nonlocal model [10] that incorporates dilatancy as a model variable, and found good agreement. This effort is ongoing, and will continue into the remaining years of his project. More details are provided in IFPRI ARR-106-03.

Paper I has been published in Powder Technology and is related to year 1 deliverables. https://doi.org/10.1016/j.powtec.2021.01.056

Year 1 Deliverables

• First statistical correlations between the various powder characteristics and their wettability and reconstitutability.
• Powder classification according to their reconstitution behavior;

The first year of the PhD work dealt with the systematic physicochemical analysis of powders and their classification according to their reconstitutability. It was achieved at the end of January 2020 with the following deliverables:

This part, focused on the impact of powder surface composition was continued during the 4th year of IFPRI project (Feb. 22 – Jan. 23) by a research engineer working on the project.

Year 4 Deliverables

• Effect of surface modifiers (quantities to cover the surface, distribution at the particle surface, minimal quantity necessary to improve wetting, etc.).
• Surface chemical mapping and nanoindentation to establish correlations with powder wettability;

The second year of the PhD work was focused on a powder presenting a low wettability (i.e. whey protein powder), which was coated with sugars to improve its wetting behavior. In agreement with IFPRI partners, five sugars (i.e. sucrose, lactose, glucose, fructose, and galactose) were chosen for their wide range of physicochemical properties: solubility, chain length, structure, glass transition temperature, hydrophilicity, etc. Links between powder wetting and sugar nature, surface modification, quantity, and coating depth were thoroughly investigated. The deliverables were achieved at the end of July 2021:

https://doi.org/10.1016/j.ces.2022.117440

Paper II has been published in Chemical Engineering Science and is related to year 3 deliverables.

Year 3 Deliverables

• Empirical models able to predict reconstitution times from powder physicochemical characteristics.
• Fitting of reconstitution kinetics followed by granulometry;

The third year of the PhD work was dedicated to modelling. On one hand, a new approach of descriptive modelling of food powders reconstitution kinetics followed by granulometry was investigated. The developed model allows to describe the different reconstitution steps by successive first-order indicial responses, thus allowing to calculate characteristic times and rates of each step. On the other hand, it was tried to develop a predictive model for reconstitution times in reconstitution conditions employed in year 1 based on physicochemical properties of powder characterized in year 1. The deliverables were achieved in November 2021:

The state of the art of capillary suspensions was described in a review paper published in Current Opinion in Colloid & Interface Science [1], reprinted below in Section A. Our model system for capillary suspensions consists of fairly monodisperse, spherical, silica particles fluorescently labeled with rhodamine B isothiocyanate in a mixture of 1,2-cyclohexane dicarboxylic acid diisononyl ester (Hexamoll DINCH) and n-dodecane, with added aqueous glycerol. Capillary suspensions are prepared using a two-step ultrasonication: an emulsification of both liquids after which the particles are added and sonicated again. The three components are all index matched and the silica contact angle can be modified [2]. The attractive interaction strength can be modified by tuning the contact angle, and fraction of secondary liquid, and, by changing the particle roughness. This lets us access both granular-like systems with weak interactions and strong attractive gels using the same model system.

During the project, we switched from using porous silica particles to nonporous particles as there were problems noted with adsorption of the secondary liquid into the pores. This means the particles are detected on the images as red rings rather than filled circles, but the particle detection algorithm based on edge detection and Hough transform is able to detect both cases. The algorithm now includes a simple graphical user interface for both local detection and manual addition or removal of missing or misdetected particles to improve the final detection efficiency. Using consecutive images on the rheoconfocal, the displacement of the particles can be detected as a function of the applied shear.

Our initial goal of the project was to track microstructural changes in the network in response to external shear applied via a linear shear cell. These structural changes were correlated with the rheological response of the material. Application of external shear via the linear shear cell, however, was unsuitable. Due to the very low yield strain in capillary suspensions, the applied shear was often above the flow point and specific changes during yielding could not be adequately captured. Furthermore, the prior setup only allowed for the deformation profile to be captured in one shear plane.

While this has provided valuable information, proving that capillary suspensions tend to undergo solid-body movement, where the rotation of particles around their respective bridges is resisted through both the structure of the network and the extra torque provided by the contact angle pinning and/or the contact angle hysteresis, full 3D tracking is necessary. Therefore, a rheometer was mounted onto a highspeed confocal microscope in 2022. The improved setup has allowed us to directly compare bulk, rheological changes with local, microscopic changes to the clusters and network, as discussed further in Section 2.3.

This project sought to develop physically realistic models for atomization processes relevant to particle production, such as in spray-drying processes, with a focus on high viscosity and non-Newtonian fluid atomization. The goals of this work were to generate a spray database and to develop understanding and correlations for the accurate pilot-to-production scaleups. We divided the work to focus on two nozzle types: pressure-swirl, and two-fluid nozzles.