Welcome, everybody, to the second course part on encapsulation by spray drying. So the second part will deal with the following topics after again, some introduction based on the S-Pro square Scheme, so we will enter into spray drying process conditions for encapsulation of function food components going into some functional details. Dynamic membrane technology to tailor make emulsion based capsules, so in order to have a well-defined capsule structure so we will couple the spray drawing with these innovative technology of dynamic membrane technologies area. And the fourth subchapter, we will enter into structure preserving stress controlled spraying, so that means gentle spraying to keep the structures which we have generated before, and then summarise and conclude.
Again, some nomenclature abbreviations at the beginning that you can work with the scriptum, and as usual, S-Pro square process makes structure, structure codes properties. We are again certainly focusing on food properties of relevance for the consumer. Now we focus on functionalities in a sensory way, so aroma flavour is most interesting, but also the nutritious health supporting functions of relevance for performance for the development of the human body from the baby, from the infant, to the adult and so forth. So is of interest in this context when we speak about functionalities.
Structure again hierarchic, so as already pronounced in the first part of this course on spray drawing for encapsulation, but now more focusing on the steps functional molecules embedded in subcapsules, so the molecules, the micro scale, sub capsules mesoscale, and they may be embedded in other larger capsules. And so if you want to have these hierarchy capsule structures, so the membrane processing as mentioned will guide us to some extent. The spray process now a bit more focusing on the tailored encapsulation based on emulsions because we can better tailor if we can make emulsion or multiple emulsion before, which can then be spray dried or also be spray frozen, so I will address both parts.
The spray drying process just for a short repetition, we combine the spraying of a liquid with some hot air contact for the drying, and the particles on that track through the spray drying towers, so they are getting rid of the water and form or developing from droplets to solid particles. Advantages of such a process shortly named here, the particle size distribution can be played with depending on the processing parameters. The moisture content can be adjusted certainly for storage and also with respect to some structural changes or preservations. Heat sensitive products can be treated because the evaporation of the water can cool the surfaces. So you can have a cool surface even though the surrounding gas is pretty high in temperature.
And we have high capacities, high throughputs such spray towers can go up in milk industry up to 30 tonnes an hour lead production. There is limitations. It’s a very energy intensive operation. Design is not trivial. Design and scale up, because the interplay of materials and the complex process and the adjustment of the final properties of the powders, in particular when there is complex structure and functionalities to be generated and preserved is certainly also not trivial, and we have to learn about the limitations in order to tackle them. One way to tackle the limitations is also using simulation tools, so we will go again, a bit further into some simulation details, time and cost reduction, process design, operational conditions, quality enhancement.
This is the idea to also support this and reach the goals by simulation tools. If we address the food in particular, so what is mainly of interest, we want to reach powder structures, which have a good flow ability, means no stickiness, no lumping. We want to have a good heat stability of these powder products, we want to have forming properties if we use these powders for products like ice cream and other dessert type of let’s say foam structures, or irrated structures, and finally, a big, let’s say, group is instant powders. This starts from infant formula and goes to instant drinks, instant coffee, soups, and so forth, where again stickiness and the nice dispersing, quick wetting, and dissolution is of interest.
Again, the interplay of a process has over the major parameters, we want to have a certain particle size distribution, certain morphology, we have certain governing parameters of the process, means the heat transfer, the temperature distribution, the heat transfer with the dimensions number, Nusselt number, characterised here with the dimensions number at Sherwood. We characterise the mass transport and the relative humidity for the exchange of the water with the air in the spray tower is very important. You have learned in another course about a Mollier diagram to give you an idea about the capacity, the take up capacity, of the air with respect to the water to be taken up.
So the spray experiments and computational fluid dynamics, we have already addressed in the first course part, allows us to even particle track and get some kinetics, drying kinetics models. We have some limitations, coalescence and agglomeration of droplets, which the spray has generated but which may, let’s say, coarsen the structure again, if we do not manage to avoid this, and certainly learning about the drying kinetics for the complex type of materials we are using is certainly always a bit of a new exercise when you enter into new recipes, new materials.
So from single drop experiments via the drying kinetics analysis to the CFD simulations, this is again demonstrating a bit the toolbox, which Loredana Malafronte has, again, used in a piece of work for her PHD thesis. Mostly we are interested to start with the very basics in order to learn about heat and mass transfer in spray drying or spray freezing and starting with single droplets is most convenient. So these ultrasonic levitater allows us to have a look at the droplets and at the same time also enter into some convective drying, and what was of interest for us was not only convective drying but also superimpose other energy input possibilities.
But first of all, what is known from particle capsule morphology during the drawing, so we may have, depending on the recipes, void formation. The temperature plays a role, whether we have different morphologies developed. The voids– or let’s say, the hollow shell type of system– can also be interesting, having a crust and a hollow inside for encapsulation. So this is known phenomenology. However the question is, is there systematics, and can we tune the systematics? For this, we entered into with the Paul Scherrer Institute, which is an ETH part having some neutron scattering experiments available. So there is a spallation source so we have high intensity newton scattering.
And so we decided to look at the single drop experiments and look at the morphology of such droplets during the drying experiments. And you can see for different dairy-based systems– for example, this is a whey protein isolate. So we can see how the droplets are shrinking. So they are getting a nonregular structure. But we can also see when we look, this would be an optical inside. And when we go for the neutron experiments, we can see the cavities forming. We can see inner voids forming during this process. Also, maltodextrins is a nice model type of system where we can see the wrinkling and the shrinkage.
And so this is best suitable, let’s say, to develop modelling rules as a function of the composition and the material. Certainly we want to have the best, let’s say, boundary conditions for forming specific structures. And this is why we said we don’t want to just go with simple fluids into the spray drying process for encapsulation, but with, let’s say, prestructured systems like emulsions.
We can look at membrane technologies to form well-defined structures to be sprayed right or spray frozen. What you can see here is a pore. Rectangular to it, we have the flow. Over the membrane, this is a membrane pore, and we see how the droplets are forming the velocity fields. So depending on the viscosity ratio, we have different shape formations and different detachments from the pore surface by the surrounding flow. This can be done in a wide range of viscosity ratios.
So such a membrane is then composed in a way, as we can see in this little video here. So we do have some droplets forming at the pore. So we can see there is not only a this kind of pore diameter, which is well-defined, but also the pore distance. This is micro-engineered membranes. So they are based on polysilicon and etched. And so we can run them also with different coatings in order to have different wetting, or also supporting the detachment behaviour of the drops from the surface. With this, we can generate simple emulsions or multiple emulsions.
We have arranged this type of membranes at rotating cylinder surfaces in order to have a well-defined flow generated on top of such rotating membranes in a concentric cylinder gap. And this allows us to adjust the wall shear stresses and the wall shear rates which are responsible for the detachment of these drops from the membrane surface from the membrane pool. This can be done with, let’s say, more macro membranes on the several micron scale, 1 to 50 micron– but also with membranes which are on the sub micron scale, as shown in the PhD work of Sebastian Holzapfel and also patented in the year 2012.
So when we look at the results from such experiments, we can easily show that the size of the droplets, of the emulsion droplets generated, depends on the wall shear stress. So increasing shear stress makes smaller droplets be generated. So we certainly want to work in a domain where they don’t touch each other, and accordingly do not coalesce with each other. This works for different continuous phase viscosities. Along the same line, wall shear stress dominates the detachment behaviour. At the same time, we can control the standard deviation of the distribution from mono-dispersed to some wider range. So as soon as we allow for some coalescence, the size ranges a bit wider when we don’t allow for the touching.
So it should work in a close to mono-dispersed manner. This simple power load equation allows us to describe, let’s say, the dimensionless drop size generated from the wall shear stress. Tau is denoting this wall shear stress. Now if we further generalise– that means using different pore diameters– and make the x-axis of the wall shear stress dimensionless, which is the capillary number at the wall, and have a dimensionless diameter of the droplets generated, which is the drop diameter divided by the pore diameter, so we can find a narrow band let’s say of dimensionless drop size versus capillary number, which again follows a power law, and which is a very nice rule for upscaling of such membrane devices from lab scale to industrial scale as has been approved in the meantime.
Results from a structural perspective. So we can see here these type of, let’s say, multiple emotions. Because we cannot only send some liquid through the pore, but this could already be a first step of emulsified system, or even already be a double emulsion. So in order to generate type of capsule in the capsule of the capsule systems for, let’s say, multiple components, encapsulation and preservation of material. If you look at the pendant, because in parallel you play with micro-fluidics devices, you can certainly even more or better tailor-make with one internal core, two cores, here four cores. So you can really play with this.
And in each of these cores, there could be another component being protected from interaction or activated in a specific state. The idea is then to fill these cores with certain functional components. What I am showing you here is micro capsules of that type being inserted into a food matrix. The food matrix in this case is a rice matrix, and into the rice kernels, which is reconstituted rice kernels from rice flour, you can have these types of capsules with sub capsules. In the sub capsules, we have iron compounds. In the surrounding, which is lipid capsules, we may have vitamins like vitamin A and antioxidants.
Because in the vicinity of the intestinal mucosa, when we can have an anti oxidative type of surrounding by release of the antioxidants first, so we have a better bioavailability of the iron, just as an example for a typical interaction of different components, which may have a physiologically relevant interaction or such preparation or type of, let’s say, conditions fulfilled for a better release and uptake. Now if we go with these complex systems into spray, we have to take into account that during spray we have elongational stresses acting besides the shear stresses.
And in particular, when we have concentrated emulsions, we can see that the apparent elongational viscosity versus elongation rate has this increase in a certain range of shear rates and increase in the elongation of viscosity. And this means we have a dilate in the shear thickening behaviour which will immediately, let’s say, modify the spray or the flow through the spray nozzle– modify the spray, and could even block the nozzle and lead to completely different structures. So this is something which we have to avoid because this would lead to a rupture of internal structures. When this was investigated by B. Dubey in his PhD work in 2013-14.
So we saw that if we look at a dimensionless scale for the size of a subcapsule, this would be the blue one here. And on a normalised scale, if we keep the value 1, so these internal subcapsules, the blue ones, would not change in morphology. And as soon as these curves go down, we have a change and a destruction of the structure. What people have found is that these curves can be shifted because what you see here. So the lambda is the viscosity ratio of the outer fluid phase, and the dispersed phase, which is taken into account, which is the yellow one with some impact on the blue one.
And when we construct a master curve by shift in a way that we modify the Weber numbers. And now we have the Weber number divided by this viscosity ratio so we can see that the curve shift on top of each other, and we have again kind of a bent width which is rather narrow in order to evaluate or to clearly see how big the We numbers, or the modified We numbers, can be to not destroy the structure. So here, they have to be below 5, which is pretty low. Because in this domain, you typically would not spray. Atomization happens at We numbers which is around 1,000 and higher.
So here we are in a domain where we are in what we call from the spray characteristics in a Rayleigh breakup domain. And the Rayleigh breakup is a filament which breaks up very regularly under very gentle relative flow conditions of the surrounding air. Related nozzle, rotating nozzle was constructed in order to generate these filaments, and generate a narrow size distribution of these types of drops. OK. Let’s have a look to an experiment with these rotational spray nozzles. So we can see in this figure how the spray is coming in. And on the left hand side, you see a sketch of these curvature of the spray. It comes in and you can see how the Rayleigh instabilities are developing.
One can easily see how you get a wavy type of surface. Then it breaks up into the droplets. So the force balance tells us that there is tension within this type of strand or filament. Centrifugal force is acting certainly due to the rotation Coriolis, because we have this acceleration in the rotational field. And from the force balance, we can then get also further information about what is governing the breakup. So mostly this is the inertia which is acting. We can also have a time series of these strands besides each other. And we can nicely see. So the age of the strand is developing from the right to the left.
And we can see how the wavy surface is starting to develop the droplets, the Rayleigh break up. This can be also modelled or described with the dimensions number, the Rossby number, in order to have the disturbing velocities and the acceleration, the rotational acceleration, taken into account. When we look at the results, so for the broken up strands filaments, we can see that here for the RPMs of these pressure rotational nozzles, we can see we are in a range between, let’s say, several hundred, let’s say 100 to roughly 500.
Micrometres, which is rather coarse but nice for capsules, because we want to have them in such a range in order to have embedded subcapsules and everything not destroyed, but keep the integrity of such systems. If we would have gone into a high We numbers spray and had atomized, this would have been the result. For the same type of system, We numbers larger than 2000 gas drop We numbers, and then we would have come down certainly to 10, 20 micron droplets. Mean size in the range of 20, 30 microns, or up to 50– but much smaller. So this would have destroyed internal subcapsule structures and so forth. This is exactly what we did not want to risk.
So the gentle spray with the larger drops and larger particles resulting, but with intact internal structure. That’s what we want to have. Then the morphologies. The morphologies you can generate in that way is different types of encapsulated capsule in capsule systems from the multiple emulsions. There could also be some bubbles inserted. There could also be in the cold, freezing tower version. So there could have been crystallisation in a well-defined way to help structuring the internal morphology, and have also glassy structures if you freeze down to very low temperatures and very quickly. So we have a large variety. And last but not least, it has to be tested with respect to functionality release.
So the controlled release of an iron compound follows a kinetic equation shown here. And just the modification of the size distributions of the subcapsules and also of the primary capsules, this is a big selection, let’s say, of these possible size distributions which can be generated. And I just take one out– so the red one– to show that this has the curve, which is also marked in red here for the kinetics, showing the kinetics, and some other ones which have, let’s say, lower interfacial area, or volume-specific interfacial area of the subcapsules where you have a slower release of encapsulated iron compounds.
So when you look at the mass transfer coefficients for the iron– Kfe, Fe for the iron, you can see the variations which we have applied here allow us to modify this coefficient by a factor of roughly 120, which is quite a wide range, just by adjusting the size distribution of these capsules. With this, the proof has been given that we can play with the structure, we can preserve the structure. We can tune the structure with respect to a controlled release. And with this, I would like to just quickly summarise. Multiple emulsions can be tailor made.
Disperse phase structure during the drawing can be preserved by the gentle drawing and controlled stresses or low stresses in the Rayleigh breakup domain, even though there is strain hardening which has to be considered as a elongational rheological characteristics of these systems. Then we can spray even double emulsions, multiple emulsions, and preserve the structure, and have the chance to, let’s say, the component released to be modified, to be triggered, to be tuned in a wide range. The Rayleigh disturbances can act, let’s say, also as a tracer for the motion of these filaments, but are in principle mainly considered in our case for tuning the size of these capsules. Rayleigh breakup has been cultivated in new types of nozzles.
And the mathematical modelling of these Rayleigh breakups was also shown to be possible with a very good agreement. With this, I would again enter into my last slide for saying thank you for all those who have contributed to this, including our funding partners funding bodies amongst which there is also Deutsche Forschungsgemeinschaft in this case. And finally, last but not least, thanks very much for your attention.