22 May 2015 The science of drying.
I've just come back from a few days at the Karlsruhe Institute of Technology's 7th Short Course on Coating and Drying of thin films. I taught a session on using hands-on computer models and apps for optimsing adhesive coatings. There was a big range of other topics from other industry experts and I learned a lot from them and from the attendees from a large variety of industries.
Although I'm familiar with the science of the drying of thin films and have written the Drying model in TopCoat, I've never had a formal education in the subject. So it was particularly interesting to get a good grounding in the subject from KIT. Historically they have been leaders in the field both theoretically and experimentally. This strong tradition continues and I was particularly thrilled to see the state-of-the-art in four areas.
- They have a super-elegant way of measuring heat transfer coefficients from experimental drying nozzles. Most of us have to rely on ad-hoc techniques for understanding how the air flows out of drying nozzles translate into the transfer of heat into a wet coating. The KIT team use sheets of thermochromic liquid crystal to both visually see the heat transfer and, via video analysis, to calculate the heat flows and transfer coefficients across the array of nozzles.
- They are serious about finding ways to map the average heat transfer coefficient throughout a dryer by using tiny thermal probes of known mass and heat capacity and monitoring their temperature as they go through the oven attached to the web. We can all hope that such devices will be commercialised sooner rather than later. One of the attendees mentioned that they had developed a similar device that produces a WiFi data signal going through the oven. We all want one!
- For those of us who are drying discrete areas rather than continuous films, they have a great analysis of why drying is so much faster near the advancing edge (there is less of a stagnant boundry layer) which can be a significant effect for those who were hoping for very even drying. And they have an elegant way (via laser interferometry) of measuring the drying of thin films under controlled air flows so that these edge phenomena can be studied in detail. Those who know of the 'coffee ring' effect will have some idea of the phenomenon, but their analysis is deeper than the popular understanding of what is going on.
- My favourite, is that they can measure diffusion coefficients in drying polymer films. Who cares? We all should! There are two drying regimes for coatings. The "constant" rate is limited by the speed at which solvent can be removed from the surface - airflows and temperature. This is usually a desirable fast process. The "falling rate" or "diffusion-limited" mode is when the drying is limited by the speed at which the solvent can diffuse through the solidifying coating. This can be a huge problem (it is commonly observed that the last few % of solvent is hard to remove) because the diffusion coefficient can fall by a factor of 1000 as the solvent leaves the polymer which then forms a more compact layer. Those not familiar with the issue of "concentration dependent diffusion coefficients" can find out more in Practical Solubility app.
This last item needs further discussion. The constant rate mode is relatively easy to model accurately (given a valid heat transfer coefficient and basic solvent parameters). It is also not too hard to model the diffusion-limited drying (such a model is part of TopCoat's Drying modeller). It just requires three parameters: a general diffusion coefficient, a parameter that gives an exponential dependence on concentration, and a parameter that gives an exponential dependence on temperature. Unfortunately, getting hold of these parameters, even for a single solvent in a polymer, is very hard. Many of us, therefore, don't seriously model the process, which is unfortunate because many of our drying problems (especially residual solvents) arise because of the diffusion limit. KIT have a beautiful confocal Raman system that can measure the concentrations of solvents (even solvent blends) throughout the drying film and from these concentrations they can find the three key parameters. How much easier life would be for all of us if we had our own confocal Ramans to do the measurements. But as they are wildly expensive we either have to use gravimetric methods (cruder and time consuming) or go to KIT for the data (highly recommended).
Because I was in the presence of true experts I could also get their views on some topics that have caused me endless arguments with those who have strong, but unscientific, opinions on drying.
- Infrared Lots of people talk about "infrared drying" and I always say "No, infrared can only heat, it needs airflow to dry". It was good to hear the KIT experts always refer to "infrared heating". We all agree that carefully chosen infrared heating can make a positive contribution to the drying process. The argument is with people who think that they don't have to have lots of airflow when they install an "infrared dryer". They think that the heat can do the drying, but drying is removal of solvent which can only be done through airflow. Infrared should be used when you know you want a quick boost of heat, for example to raise the temperature of a thick web or to transition swiftly into the higher temperatures needed for diffusion-limited drying.
- More air! The KIT experts have (sadly) seen what I have seen. Those with lots of residual solvents think that they can solve the problem by blasting the final zones with lots of air "to bring out the solvent." This is an expensive way to achieve nothing. Air has to be maximised in the constant zone (though not so much as to disturb the coating, which can be managed by by good nozzle design and by ensuring high viscosity as soon as possible, which also reduces Marangoni effects), but the only way to speed things up in the diffusion zone is to raise the temperature to the maximum possible without harming the coating via other effects. This is because the diffusion coefficient has a strong exponential dependence. At the very least, a temperature rise of 10°C will double the diffusion coefficient and halve the time to remove the solvent. So a good oven setup directs most of the airflow to the early zones and in the later zones uses only enough air to ensure rapid heating of the web to the new temperature and safe, low, concentrations of the solvent in the air.
- Skinning It is common to hear that "skinning" is a problem in drying along with a solution to the problem which is to "dry from below". The folk physics is that the skin is produced by too much drying on the top so drying from below will fix the problem. I was delighted that KIT meet this folk physics too and have to point out why it is wrong. Drying can only take place from the surface - drying is removal of solvent. Whether heat is supplied from the top or bottom makes no difference to this. In addition, the temperature difference from top to bottom of a coating (in fact, from the top of a coating to the bottom of the base film) is very small. The fact is that all coatings "skin", i.e. they all reach a diffusion-limited region. A good formulation keeps the polymer/ink open for as long as possible (usually via good polymer/solvent compatibility which in turn can be optimized via Hansen Solubility Parameters) so that diffusion is as fast as possible. And a good drying setup makes sure that the temperature is never above the BPt of the solvent if/when there is a considerable amount trapped by the diffusion limit. This is because people aren't so much worried by skin as by blisters caused by over-heating of diffusion-limited coatings that still contain a large % of solvent. By understanding that temperatures are low during constant mode (the enthalpy of vapourisation cools the coating) and high in diffusion mode, and by knowing where that transition occurs, skinning can be controlled rationally. "Heating from below" is not a rational way to fix the problem.
- I admit it, I used to have no idea when my coatings made the transition from constant to diffusion-limited mode. So my oven settings were far from optimal. But once I got into the habit of knowing where the transition happened (steep temperature rise and/or touch dry) it became much easier to optimize the process. First, I could get a good estimate of the heat transfer coefficient of the drier, using the TopCoat model and varying the HTCF input value till there was a match between predicted and experimental transition point. Second, by getting much faster drying of temperature-sensitive coatings by setting much higher temperatures in the constant zones which (from the TopCoat model) would never come close to the temperature limit. Third, using TopCoat's "Blister" warning, I could see if the residual solvent in diffusion mode was in danger of reaching temperatures above the boiling point. To get the transition data I had to rely on relatively crude techniques with IR guns, opening the oven to look for the touch-dry point, or by looking at the temperature at the end of each zone with a built-in IR guage. The point here is that KIT strongly recommend (and I agree!) having an accurate temperature map throughout the oven. But they also agree that having a crude map is much better than the alternative which is having no map at all. This alternative, unfortunately, is far too common.
My thanks again to the KIT team for a great few days of coating and drying science. I have, of necessity, simplified their views in this blog. If you want to find out what they really think, please contact Professors Wilhelm Schabel and Philip Scharfer.