The optimisation of freeze drying processes

From its infancy to today, lyophilisation has seen a constant but very slow evolution. Indeed, the techniques that we utilise today differ very little to those used industrially, after its development as a viable technique for the preservation of serum during World War II. Its application to pharmaceuticals and from there on to foods at commercial scales followed in very short order, becoming pervasive through the 1960s and 1970s to almost exemplify human progress in people’s imagination as the food of the space age in the 1980s.

In fact, it would be tempting to say that almost everything that can be invented in freeze drying has been invented, and that any small tweaks that developers today can think of are merely that: tweaks. Where, then, does that leave the technologist or engineer faced with the problem of reducing costs in the very expensive world of freeze drying? In a world where commodity, energy and food prices soar and where our consumers are ever more aware of the environmental impact of what they consume, the competition for the most cost effective and sustainable processes is won or lost in the ability to make those tweaks and to successfully implement them, and the full understanding of the science of the process becomes ever more important.

So what is the final objective of this battle? The first important thing to understand is that freeze drying is a tremendously capital intensive process. To give an idea, the cost of a standard industrial freeze drying plant may run to tens of millions of dollars. In that context, making the most of the freeze drying assets that one has already bought is a great way of having to spend the further tens of millions of dollars in new plants, money that can then be invested to improve the consumer’s experience, or taking a price out of the final product. It also requires a large amount of energy to be effective. Therefore, reducing the amount of energy required to run it will have beneficial economic impact, but more importantly help us reduce carbon emissions.

For a long time freeze drying has been understood as a blend of art and science, where understanding the molecular details of the process is meaningless when overlaid with the massive complexity of running a factory with tens of thousands of tonnes per year capacity. It is very tempting in those circumstances to see the problem of keeping a productive and competitive factor as completely divorced from that of understanding the fundamental science behind what is going on. That is of course simply not the case. A pragmatic approach to bridge the gap between art and science in this, among many others, industrial process has been the application of computing power together with multivariate statistic techniques to correlate the impact of variables that one can measure and to some degree control with the final output, allowing us to understand what variables to control in order to obtain the desired output and the development of models that can help us calculate how to increase throughput or reduce energy.

This approach is living a moment of great popularity throughout the industry under the name Six Sigma. Lean Six Sigma offers tools to putting this in practice that are being extremely successful in the industry. The application of these can allow us to obtain key scientific information at the same time as it allows us to pragmatically adjust variables to achieve the desired outputs and are as likely to accelerate progress in improving freeze drying at a very fast pace.

However, the blind application of such processes also has some limitations. For example, the choice of the wrong variables in the exercise may result in misleading information. In other words, one may attribute a causal link between a variable or group of variables and an outcome where no such link exists, leading the engineer or technologist to choose the wrong path for optimisation.

The other limitation is that, in order for these processes to be effective, they require large numbers of trials on an industrial scale, which by their very nature may end up resulting in large amounts of products that one will have trouble dealing with as they are out of an agreed specification. While most companies are likely to accept that this is a worthwhile risk to take, it will also limit the amount of experimentation that is possible.

Additionally, as it is very difficult in these sort of studies to extrapolate (that is, to reach conclusions outside of the limits that have been studied in the trial), it means that it is very difficult to take the learnings.

The solution to this is to apply an approach that uses the power of Six Sigma type experimental design and blends it with scientific understanding. By understanding not only what effect key variables are having on our product but also why, we are able to expand our knowledge beyond the reach of our experiments, transfer learnings from different experiments and effectively leverage pilot or laboratory scale experiments.

So what are the fundamental parameters that one must understand to be able to control costs in the freeze drying process, and are we so far away from being able to implement detailed scientific knowledge at commercial scales?

To answer those questions, we must go back to defining the freeze drying process and breaking it down into three fundamental areas. For the following, I will take the example of a viscous fluid that is dried from a liquid solution of high concentration to dryness.

Freezing

In this stage, our viscous fluid is cooled below the temperature where it is completely solid. At this stage in the operation, the frozen slab may be modified to reduce its particle size, and provisions must be made to account for the frozen slab gaining some heat in this process (i.e., we may have to freeze the sample to a lower temperature than what otherwise may be necessary).

During this stage, free water will start to freeze bellow 0°C, and form crystals. The speed at which this occurs is critical; if enough time passes while the crystals are grown, at the right temperature, smaller crystals will start to aggregate to ‘build’ larger crystals with dendritic forms (the word dendrite comes from the Greek, meaning tree, and are so called due to their treelike shape) in a process known as Ostwald ripening. However, if the freezing process occurs very rapidly, such a process will not occur and the crystals will be significantly smaller.

Another important consideration is how much of the water will actually crystallise. In order to understand this, it is critical to differentiate free water from bound water (which is bound by a strong intermolecular force, hydrogen bonding, to molecules of the foodstuff that contain hydroxyl residues) of which only free water can crystallise.

In freeze drying, it is generally desirable that the maximum freeze concentration (that is, the largest possible amount of crystals) is achieved, for reasons that we will discuss later. To achieve this, the technologist can use a number of tricks; for example a technique named annealing. Annealing is a process that is well described in literature and extensively utilised in freeze drying, which takes inspiration from the age old tempering process in metallurgy. In this process, the frozen slab is melted under controlled conditions, by increasing the temperature slightly, and then cooled again. This process is repeated a number of times so that ice crystals are formed and then recrystallised in their most stable form, and thus the maximum freeze concentration is achieved.

We will now direct our attention to the second stage of freeze drying, where it will become apparent why the freezing process is so critical.

Sublimation

Sublimation is a phenomenon by which a solid may under certain conditions go into a gaseous form without going through a liquid phase. Below 6.11 mbar, all ice is converted to gas without going through a liquid phase, which is the reason why freeze drying is always conducted under vacuum. In order to optimise the capacity of a drier, as well as to make the process more energy efficient, heat applied in this phase must be invested in ‘paying’ the latent energy of sublimation for this transition as efficiently as possible.

In order for that to happen, one must be able to transfer all the mass of vapour generated during sublimation to condensers in the freeze drier as efficiently and fast as possible; this will determine the speed at which this phase of freeze drying may be conducted.

As we discussed earlier, the physical structure of the ice slab at this stage has been largely determined by the amount of free water available and how we crystallised that free water. If large, dendritic crystals have been formed, these will sublimate and leave large dendritic cavities for the rest of the vapour to move through, overall increasing the efficiency at which that happens. So by investing time and energy in the freezing process one may save time and energy at the sublimation stage.

Desorption

Once all the free water has sublimated, the water that is still bound by hydroxyl containing molecules will desorb. For this to happen, an increase in product temperature sufficient to supply the energy required for desorption is necessary.

At this stage, again the technologist or engineer must take special care, as by excessively increasing the product temperature the slab may transition from its glassy state to a rubbery stage and collapse. Again, then, the secret here is to maintain balance: Between sufficiently increasing the temperature to enable desorption of bound water without increasing it to the point where the material collapses leaving behind a sort of sticky goo.

By carefully managing this stage, it is possible also to control the final moisture of the product within very tightly set control parameters that will ensure the final product does not stick or cake and maintain all its desirable sensorial attributes.

In conclusion; the optimisation of freeze drying processes to allow capacity improvements and energy efficiency is an ancient balancing act that requires detailed scientific understanding not only of how the process operates but also of why, and can be significantly accelerated by collaboration of professionals in laboratories, pilot and commercial plants thanks to the statistical packages that are gaining in popularity today.