Developing a Transferrable Lyophilization Protocol using a MicroFD with LyoPAT
February 27, 2020
Presenter: Spencer Holmes, Application Engineer, Millrock Technology, Kingston, NY
This webinar demonstrates how the MicroFD with LyoPAT can be used to both develop an optimized lyophilization protocol and then transfer the protocol to a pilot or production scale system.
Hooke College of Applied Sciences, a member of The McCrone Group, occasionally offers a 2 ½ day Lyophilization Process Development Workshop. Contact us for details.
Charles Zona (CZ): Okay, I think we’re ready to get started. I’d like to welcome everyone to another McCrone group webinar. My name is Charles Zona, and today we are happy to welcome Spencer Holmes from Millrock technology.
Spencer is going to talk to us today about Developing an Optimized Transferable Lyophilization Cycle Using a MicroFD with LyoPAT.
But before we get started, I would like to give you a bit of Spencer’s background. Spencer is an applications engineer at Millrock technology where he works with the engineering team to continually develop and improve new tools for both analyzing the thermodynamics of freeze-drying, and optimizing lyophilization protocols.
Spencer’s main area focus is the MicroFD and the LyoPAT, and he also instructs lyophilizer operators on how to use these technologies to analyze and optimize their lyophilization cycles. Spencer is also one of the instructors for Hooke College’s lyophilization course.
Spencer will field questions from the audience immediately following today’s presentation.
This webinar is being recorded and will be available on the McCrone Group website under the Webinars tab, and now I will hand the program over to Spencer.
Spencer Holmes (SH): Hi, thanks for the introduction, Chuck. So the topic of today’s webinar is going to be Developing an Optimized Transferable Lyophilization Cycle Using a MicroFD with LyoPAT. A couple objectives when we’re looking at developing a new freeze drying cycle or optimizing an existing one is that we want to be able to do this using a minimum amount of valuable active pharmaceutical ingredient or protein product when we’re developing the cycle.
What we need to do throughout the cycles—we need to measure and calculate our critical process parameters such as our heat and mass flow, our kV bio heat transfer coefficient, and our product cake resistance, RP. By developing, by determining these critical process parameters, you can develop an optimized product thermal history and use this to develop a robust transferable protocol.
The piece of equipment that we will be specifically talking about to do this work is the MicroFD with LyoSim and LyoPAT. The MicroFD is a small freeze dryer that can utilize between seven to sixty-one vials, depending on the size of the vials, to completely develop a freeze-drying run.
The LyoSim is a system in the MicroFD that eliminates the edge effect, making all the vials behave like center vials. Essentially, the LyoSim is the piece of equipment in the MicroFD that makes running a small batch of vials possible or feasible. This can’t be done, for example, on a larger tray unit by just using a small array of vials, because that will not be representative of what a full tray will behave like, so when we’re using the MicroFD with the LyoSim, it eliminates the edge effect, so if vials behave like center vials, and we can use this as a representative cycle for a larger freeze dryer.
The quick theory behind this lies behind this LyoSim, is that based on the edge effect, as many are familiar with, is that vials on the outside of an array are only in contact with three or four other vials—three or four other points of contact—and even not just the very outside rows, but the second row, and to some extent, sometimes even the third row also experiences some form of edge effect due to the effect of the convective energy on the outside of a bottle. So the solution is to bring in temperature controlled LyoSim blocks that simulate the presence of a wider array of vials for this small array of vials, thereby creating uniform heat flow for the edge vials of the array as if they were in center vials. So the complete design solution for this is a temperature-controlled LyoSim ring on the outside of the array, and then aluminum blocks resting on the ring that bring that temperature controlled ring into thermal contact with the array of vials when we use this LyoSim ring, and we set this ring to track the product’s temperature during the primary drying phase.
So again, these blocks on the outside are all the same temperature as the product in the vials, here, simulating again a wider array of vials present in the freeze dryer, which leads to a uniform batch of drying. Here, we can see some example test results where we have a percent dried measured gravimetrically between 23 and 26 percent, which is well within the range of uniformity you would see in a full batch of vials in a tray freeze dryer.
I’ll have just a quick, quick overview of the MicroFD in the LyoSim— the freeze dryer, the small-scale freeze dryer—and the LyoSim ring which makes using a small scale freeze dryer possible; just as important, if not moreso when we’re looking at developing and optimizing a cycle, is LyoPAT, which is a suite of advanced tools for freeze-drying which include FreezeBooster for controlled nucleation, AccuFlux for post-nucleation heat flux control, and AutoDry for closed-loop primary drying process optimization. So when we’re looking at developing a new or optimizing existing freeze drying cycle using the LyoPAT features available in MicroFD, there’s a general process that we like to like to recommend and look for.
This starts with analyzing your existing or new cycle by running a regular, plain, recipe-based freeze drying cycle, either based on the existing conditions you run at currently, or based on, typically, a more conservative looking cycle that you can safely process your product at.
During the cycle is when we can calculate, using the LyoPAT heat flux sensor, our critical process parameters throughout the run, and we’ll use our post-processing to get normalized results accounting for all the heat flow in the system.
We then can look to optimizing this run using the various optimization features at our disposal. These are again FreezeBooster for controlled nucleation, AccuFlux for direct heat flux measurement and control for post nucleation, and AutoDry for primary drying cycle optimization.
So we notice here that two out of three of these features are built based on optimizing the freezing side of the freeze drying recipe, so only AutoDry actually works in primary drying. The FreezeBooster for controlled nucleation and AccuFlux are features that are used in the freezing phase, because, really, the foundation of an optimized protocol occurs in the freezing phase. That’s when you’re developing the frozen crystal structure within your product that will then be dried off, and that will really determine a lot of the critical quality attributes of your final dried cake.
So using these three optimization features, we’re going to optimize our cycle to develop a better cake, a more uniform cake across the batch, and to reduce the cycle time as much as possible. And once we have this optimized protocol, we can then look to transferring it using the critical process parameters that we’ve calculated, and by comparing the kVs between various units that we’re looking to transfer to.
A brief overview of a case study that was done using the MicroFD of reducing a cycle time shows that broadly, we were able to reduce the cycle time by over 40% from using just, on the left here, a completely recipe based cycle, where we’re just ramping during freezing and then ramping up and drying at a steady temperature, and primary drying, and then every successive bar here represents the addition of one additional optimization feature.
So first, we have controlled nucleation, then we have controlled nucleation and the AccuFlux heat flow control post nucleation. And then finally, we have the controlled nucleation, the post-nucleation AccuFlux control, and AutoDry and primary drying. So we notice that the first two steps of the optimization here did not really lead to drastic, or even significant, reductions in primary drying time, but they did yield noticeable results in the structure of the product we are drying, and we’ll take a look at that coming up.
So again, to start this process, we started with a plain recipe-based freezing program where we just ramp to a steady temperature in freezing and primary drying, and during the cycle, LyoPAT automatically calculated all of our critical process parameters throughout this run, including our kV, our mass flow, and our cake resistance. The specific recipe used here was ramping at one seed per minute to -40 and then drying at -25°C and 60mT. The product used for this example was 5% sucrose, 2 milliliters in 6R vials, and there were 19 vials used in this array. And again, these are some of the critical process parameters that have been calculated for us during this first analyze run by the LyoPAT software using that heat flux sensor. We calculate our bile thermal conductivity, cake resistance, our mass flow, and then we have our product temperature and shelf temperature from standard thermocouples.
And these are the data results we’ll see again from our basic ramp freezing recipe. This red line we see here is the heat flux that we’re measuring from the vials, and the green line represents our product average product temperature. So we noticed that this green line has a bunch of little jagged marks in it, and each of those jagged marks represents a single vial that’s being monitored with a thermocouple, nucleating spontaneously. They also correspond to dips in the heat flux as well. As that vial nucleates and the temperature rises up close to 0°C, the temperature difference between the product and the shelf increases and we see a sharp increase—a sharp negative spike—in the heat flux.
And what we notice here is that we have vials nucleating at a range of different times and temperatures followed by a sharp deep V in the heat flux. So what this means is that due to the random spontaneous nucleation, we have non-uniformity across the batch of vials, since their nucleating at different temperatures and different times, and then after nucleation, they’re going to be freezing at different rates. We have non-uniformity across the batch of files, as well as non-uniformity within the crystal structure of each vial, as well, because once we initially nucleate, and we see it approximately, maybe 10 or 15° difference between the product temperature and the shelf temperature, and we’re freezing at approximately a rate of, for example, maybe 800 w/m2 early on in the freezing. That heat flux here of -800 w/m2 is directly proportional to the rate of freezing and crystal formation. So the faster we’re freezing, the smaller crystal structure we’re having. So initially, we’re freezing, you know, may be relatively slow at -600 or -800 w/m2, but as that shelf continues ramping down, and our product, while freezing, still stays up near the freezing poin —close to 0°C—that temperature difference begins to grow so that heat flux begins to increase in magnitude, as well, where towards the end of the freezing, we’re now freezing at close to -1200 or -400 w/m2. So we’re freezing at almost twice the rate, which means we’re growing ice crystals much quicker and much smaller—this leads to non-uniformity within the structure of each vial.
Some of the results we get after this is that for this run, again, drying at -25°C, we have a primary drying time of 26.7 hours. That doesn’t really say much, since we’re not comparing it to anything right here, but we’ll take a look at how this compares as we optimize the cycle in the following slides.
So as we mentioned before, one of the most important parts of optimizing a freeze drying cycle begins in the freezing phase. So we want to understand the impact of varying our freezing methods by either being freezing rates, the addition of an annealing step or two, or what we’re going to address more specifically: using controlled nucleation.
So, why do we want to control nucleation? There are a couple really important reasons why.
Most importantly is that it creates a more homogeneous batch by forcing crystallization to occur in all the vials at the same time, at the same degree of super cooling. This produces both consistency across that batch as well as consistency between subsequent batches that are all being nucleated at the same time. Additionally, in some cases, it is shown to reduce cycle time, but that’s not the primary purpose or benefit of controlled nucleation. That is something that happens in some cycles, but primarily, the benefit of controlled nucleation is that it produces the same degree of super cooling and increases homogeneity in batch consistency.
So here, where we’ve used controlled nucleation, we can take a look at the data results we’ve gotten from the freezing phase of using controlled nucleation. Here we can say, by looking at the green average product temperature, that we have one single large spike, which means all the vials with thermocouples are nucleating at the same exact time and at the same exact super cooling temperature.
And again, we see one sharp spike in the heat flux, but then it is still followed by that deep V. So although we’ve used FreezeBooster, we’ve used controlled nucleation to get uniformity across the batch here, we still have non-uniformity within each vial due to that varying rate of freezing during that freezing process as that shelf is ramping down. So again, early on after nucleation or freezing at a relatively slow rate of -600 w/m2 here, then as our shelf temperature continues to drop while our product temperature remains largely the same, that temperature difference grows at towards the end of freezing, we’re freezing at twice the rate of -1200 w/m2 here. So there’s still that non-uniformity within the structure of each vial, but we did create more homogeneity and batch consistency across the batch.
The results we get from this, again, we can see for this run we had a drying time of 26.1 hours, although that’s not statistically significant or drastic. What we do see is, we see an improvement in our product cake resistance and the temperature of our product as it’s drying. One of the reasons we don’t see a large reduction in primary drying time is that in this first run we had all of our vials nucleating between -3 and -5°C because these were not run in a cleanroom environment. So there was, you know, dust or particulate that led to nucleation before large degrees of super cooling. In a lab, or production cleanroom environment, we can see, typically, you know, -10, -15 or even up to -20° of super-cooling before auto nucleation, and when they nucleate at such a low temperature, it leads to rapid, very small crystal formation and a higher cake resistance. So we didn’t see a huge reduction in primary drying time between here because we were already nucleating at a relatively high temperature of -3 to -5°C.
So when we did controlled nucleation at minus-5°C, we didn’t improve that nucleation time too much, but that nucleation temperature too much or our primary drying time, but we did improve our cake structure, and we improved our product temperature. So although we continue to dry this run at -25°C, we potentially could have increased this temperature by maybe 5, or 2 -3, or maybe 5°C more while keeping our product temperature still safely below its critical temperature.
Now that we’ve used controlled nucleation as the first step of optimizing our freezing side of this freeze-drying recipe, we want to finalize and completely optimize this freeze drying recipe by using AccuFlux for post-nucleation heat flow control. What this feature does is it controls the shelf temperature based on a user-set heat flow setting, for example, -400 w/m2, which says after nucleation we want to maintain the steady heat flow for the rest of the duration of freezing, which leads to a constant rate of crystallization and a uniform crystal structure throughout the vial. One thing important to note, is, that for typical degrees of super-cooling between -5 to -10°C, during that nucleation event only about 8-12% of the water actually freezes and forms an ice crystal; the rest slowly crystallizes after that nucleation event, and that’s where we see the non-uniformity in the vial if we just straight ramp the shelf temperature down. So we want to use the AccuFlux to steadily freeze that remaining 90% the water.
And here’s what this data looks like for a run that’s using both controlled nucleation and AccuFlux. So we see again the single nucleation event where we’ve achieved batch uniformity through controlled nucleation, and we see then that we’ve eliminated this deep V of freezing by telling the system to control the shelf temperature to maintain a certain heat flux setting. For this run, the setting was -400 w/m2, so we initially had to ramp the shelf down a little bit to get the heat flux there, and then it maintains it pretty steadily there. We see it’s not a perfectly smooth heat flux line here, but it’s a vast improvement on that deep V of freezing we saw before. So we’ve both, again, achieved uniformity across the batch as well as uniformity within the structure of that vial.
And here are the results we see after this. Again, this was then after this freezing protocol. It was dried using that same conditions of – 25°C 60 mT, and we saw a slight reduction in primary drying time of about 8%, which is something, but it’s not crazy. But what we did see, which is very significant, is an over 2.5° reduction in our product temperature as it’s drying. And we’ve effectively halved our product cake resistance by creating such a uniform crystal structure throughout each of those vials. We’ve left a uniform dried cake structure, once all that ice is sublimated away. So again, why we, for this run, maintained the same primary drying conditions of -25°C, for comparison sake; we easily could have increased this shelf temperature by up to 10° more, while still keeping our product safely below its critical temperature.
Now that we’ve completely optimized freezing through both controlled nucleation and the use of AccuFlux for post nucleation heat flow control, now we can finally take a look at optimizing our primary drying. The feature we use to optimize our primary drying is called AutoDry, which is based on a closed loop feedback based on the thermocouples in your product.
So the theory that AutoDry uses is that early and primary drying, where you have no cake resistance built up, is when you can more safely maximize the heat input, or shelf temperature, to your product. Later on in the cycle, once you start building up a cake resistance, that’s going to start raising your product temperature and raising the effect of pressure at the sublimation interface. That’s when you want to then produce your shelf temperature to keep your product safe, but early on when there is no dried cake resistance and your interface pressure is the pressure that you’re controlling at, that’s when you can really drive a lot of heat into your product and all of that heat is going to go straight towards sublimation, without warming your product temperature beyond its critical temperature.
So that’s the that’s the theory that AutoDry Works on is that it’s going to maximize the shelf temperature early on in the cycle and then gradually reduce it to maintain your product below it’s critical temperature. At the same time, the AutoDry feature conducts a pressure drop test, which is used to determine any thermocouples that are out of ice. So during the pressure drop test, which happens at a user-set interval between 30 to 180 minutes, every interval, the system will reduce the chamber pressure by setting the set point to 0 and reduce the chamber pressure for a minute and a half.
And during that time, as the pressure in the chamber is reduced, it shifts the solid vapor equilibrium to a colder temperature. So all the ice remaining in the product is going to reduce in temperature, and any thermocouples that are still in ice, still submerged in the ice, are going to see a temperature drop as well.
So any thermal couples that do not see a significant temperature drop are then considered out of ice and we won’t use those for control, we’ll only use the thermocouples that are still in ice. So if a thermocouple pops out of ice early and starts rising in temperature, we know we don’t have to worry about it, because it’s not actually in your frozen product.
So the results of this AutoDry run we can see here…for this run, we used the same freezing protocol as the run before it where we used both controlled nucleation and AccuFlux for post nucleation heat flow control, and then instead of drying at a steady temperature of -25°C, we used AutoDry.
So for AutoDry, we set our baseline temperature—our initial temperature at that same -25°C here, and we set a time period of 90 minutes for it to hold at this temperature and for the thermocouples and shelf and everything to reach equilibrium. After this whole time had elapsed, we conducted our first pressure drop test, seen by these small spikes, and then we began our shelf adjustments.
We can see that our shelf temperature increased up near to about -3°C, and every hour or 90 minutes or so, it’s conducting this pressure drop test to determine if any thermocouples are out of ice, and as our product temperature begins to rise and approach our critical temperature, and then lowers the shelf temperature back down to maintain our product below that critical temperature before, finally, near the end of the cycle, shortly before all the thermocouples begin to pop out of ice it reaches a final set point temperature. The settings we used for this run: we used sucrose. So we set a critical temperature of -32°C and we set a 2° safety offset, which means that AutoDry is going to try to maintain our product at -34°C.
So we see there are sometimes, right around here for example, where it looks like the product may have approached or even surpassed that -34°C a little bit. That’s why we have this 2° safety offset. And as soon as it passes that target temperature, AutoDry rapidly ramps that shelf temperature down to reduce the product temperature and keep it safe, below our critical temperature.
Results we see from this are that our primary drying time has been reduced to 15.2 hours, so a pretty significant reduction in primary drying time. During primary drying, AutoDry brought the shelf temperature to a max temperature about -3°, and then a shelf temperature about -14°C is the final temperature it landed on. So it reached a max temperature -3, and then we reduced down to -14, and dried there, and we can see that this reduced the primary drying time by half and still kept the product safely below its critical temperature, minus that 2° offset. So again, that was -32°, the 2° safety offset we controlled at -34°C.
The final results of this run is that we’ve taken our initial recipe-based freeze-drying protocol at 26.7 hours and we’ve reduced that, the 26.7 hours of primary drying time, reduced that to 15.2 hours, which is the total reduction of about 43%.
This is what the final optimized results look like. Here, we can see the Pirani convergence for both of these, as well as the reduced optimized product peak resistance.
So as a quick summary, the steps we took to optimize this protocol were to run an existing recipe-based cycle, then to begin to optimize the freezing phase by choosing to vary freezing rates, add an annealing step, or in our case, a controlled nucleation. And then we fully optimized freezing by using both controlled nucleation and post-nucleation AccuFlux heat flow control. The final step for creating a completely optimized cycle was to enable AutoDry and primary drying, which then optimized the primary drying phase. We have both a fully-optimized freezing and a fully-optimized primary drying phase in this cycle.
Okay. So now that we’ve looked at how to develop an optimized cycle, we want to look at how we can take a cycle and transfer it to a next size unit, so we’re going from lab to pilot, from pilot to production, and the theory is essentially the same: that is, that the goal when we’re transferring is to maintain an equivalent product temperature thermal history between the lab and commercial processes.
One standard safe way of transferring a cycle is to maintain the same settings we used on our smaller scale unit and then extend primary drying time. So the trend that we see is, that the larger the unit we move to, the inherent kV of that cycle goes down. So if a cycle runs safely on a smaller unit, when we take those same settings and run them on a larger unit, we can be very confident that they will run safely—and we can see that here.
Here we have a cycle that was dried in a MicroFD at a 100mT at 0°C. In the red line, here, and the corresponding product temperature in the purple line, here, and that same recipe—this product was 5% mannitol—was then run in a Revo, or a lab scale unit, at 0°C. Then we can see the heat flux in the center of the batch for this cycle in the orange line and the corresponding product temperature in the light blue line. So here we can see that it’s drying at a slower rate. We have a lower heat flux, which means a lower sublimation rate, and overall, it takes about 30% longer to run. So again, a safe way to transfer it is just use the same cycle conditions and extend that primary drying time.
Another way to transfer cycle. Let’s take a closer look at the kVs that the files are running at, and each system. So here we take our formula for kV, and this can be determined through AccuFlux or use the heat flux sensor to directly measure the kV, or gravimetrically by weighing vials before and after primary drying during a portion of the primary drying cycle. So here we take this equation for kV and we rearrange it a little bit so we have heat flux on one side all on its own, so we have kV × Tshelf – Tproduct = heat flux. And we’ll do that for both systems, and then we’ll set those heat fluxes equal to each other, and we’ll rearrange it again to see what the new Tshelf we should use is. So if you do that, here’s the equation we get when we’ve rearranged it, and, essentially, comes down to the ratio between the kV of the source unit and the kV of the target unit that we’re transferring to. For this example, and in this form of the equation, the kV we used uses the shelf surface temperature to measure the kV—so we measure the kV using the shelf surface temperature in our source unit, and in our target unit. The same exact thing can be done if we use our kV using the shelf inlet temperature. We just want to make sure that whatever reference were using, whether it’s a surface or the inlet temperature, is consistent between each unit. So again, for this unit, we used the shelf surface temperature. We also need to include an additional term, this ΔT, which is the difference between the inlet and the surface temperature of the target unit.
Comparing our MicroFD and our Revo: in the MicroFD. We had a kV of 22.16 w/m2 C and in the Revo, in the center of the shelf, we achieve 18.01 w/m2 C. So here’s where I’d like to mention that when we’re looking to transfer a cycle to a larger system using this method, we want to focus on whether we’re transferring with the center vials in mind, or transferring with the edge vials in mind. If we’re transferring with the edge files in mind, we would use the kV of the edge, which would be a higher kV. And when we do that, we’ll transfer our cycles with our edge vials, we’ll dry the same as our vials in the MicroFD did, which may lead to our center vials taking longer to dry than it being a little conservative for the center vials, but also safe for the edge vials. At the same time, as we did in this example, we can transfer with the center vials in mind, so we’ll use the kV from the center of our batch. So when we do this, we’re going to transfer so that the center vials dry the same in the Revo as they did in the MicroFD, keeping in mind that when we’re transferring with the center in mind, which is more of an aggressive cycle, we’re going to do this so that the center vials dry the same as they did in the MicroFD, but under these conditions, the edge vials may dry a little too warmly, so it may be risking some of your edge vials if you’re transferring with the center in mind. And this comes down to a product and a product decision and the company-based financial decision. If we’re looking at a very large freeze dryer with a large tray, then those edge files may only be 5% of your total batch, and maybe it’s more efficient to then transfer with the center vials in mind so we have a faster batch drying and 95% of those center vials are drying faster and are usable.
For dry transferring to a smaller unit where the edge files are a considerable portion of the batch, maybe 20% or 30%, then we may want to transfer this cycle with the edge vials in mind, because we don’t want to risk ruining 20% or 30% of the batch. So we’ll transfer with the edge vials in mind, and the center files will still dry a little bit more conservatively, then a little bit slower, but we’ll have a full safe batch. So those are the two considerations, or one consideration with two main options. You want to keep in mind when you’re looking to transfer cycles is if you want to transfer with the center vials in mind, or with the edge vials in mind.
So back to this example. Here we were transferring with the center in mind, so we use the equation from the previous slide, comparing the kVs between the MicroFD and the Revo, and we see that the Revo has a slightly lower kV than the MicroFD. So what this tells us is, that we want—and again this kV was based on a shelf surface temperature—so the ration of the kVs here is telling us we want our surface temperature to be about 2.7° higher than it was before; we want to be about 2.7°C, accounting for that difference between the inlet and the surface temperature of 1.5°C—this gives us a control inlet temperature of about 4.2°C, for which we used 4°C.
And the results we get looking at this is that we can see that this orange dotted line, the center Revo heat flux matches the heat flux of the MicroFD almost exactly. So by comparing those kVs and adjusting the shelf temperature based on the difference in the kVs, we can achieve a virtually identical thermal profile going into those vials. We can see that the product temperature in the Revo here, when we’re drying at 4°C, is a little bit higher than the MicroFD, but we can see—for earlier on in the cycle—we can see that these end up matching up almost identically towards the end of the cycle, right before the thermocouples are popping out of ice.
Overall, we might say that this was, +4°C, was a little bit too aggressive, and maybe using 3°C instead would have been ideal, but in any case, we’re able to look at the two kVs between the source and the target unit, compare them, and adjust our shelf temperature accordingly to achieve the same thermal history result. And we can see the same thing when we look at the Pirani and capacitance convergence, which leads to a cycle time that it is virtually identical, only 3 minutes apart.
So overall, after we’ve optimized our cycle and we go to transfer it, we’ve done it in such a way that the sublimation rates and the heat transfer were very similar between the two machines, resulting in a similar, or almost identical, processing time, and throughout this process, we were able to do this using only 19 vials rather than hundreds for every step of the way. So every consecutive run between optimizing and then transferring we’ve used 19 vials in our MicroFD. Now depending on the vial size, we can be using between seven and 61 for very small vials, but still much less than the hundreds that are required to do this work on a tray-style lab development unit, and we’ve still been able to transfer in such a way that we have the same results in a production unit. Through and throughout this process, we have been able to save significant time and money and made it an enjoyable development process. So thank you. That is the end of the regular webinar here, and now I think we’ll open it up to questions. Thanks.
CZ: Okay, thanks Spencer was a great presentation. We’re going to move into the question and answer part of the webinar now, and if you have any questions, please go ahead and type them into the questions field. Looks like our first question comes from Kris. Kris wants to know, in your optimization example, controlled nucleation was used in both optimization steps. How can someone optimize their freezing recipe without controlled nucleation?
SH: Well, that’s a good question Chuck and one we hear relatively frequently. Without the ability for controlled nucleation, unfortunately, your options for optimizing the freezing side of your cycle are a bit more limited. That’s where, it’s something I briefly touched on in the presentation, you can start looking at experimenting with varying freezing rates as well as the addition of a potential annealing step. The addition of the annealing step is when you raise the temperature of your shelf and products to put a little more energy into that crystal lattice structure and allow the ice crystals to kind of rearrange into larger ice crystals, which does help improve the uniformity within your vials. Additionally, on the other end, by varying different freezing rates, if we look back at some of the plots for, like, the heat flux that we see during normal, for example, in our in our experiments, 1°/minute freezing rates, is that if we’re lowering the shelf at a rapid rate like that, although one degree doesn’t sound very rapid, we see that it does pretty rapidly outpace the temperature of the product while it’s freezing. So instead of having a consistent or steady 1° ramp down to your final temperature, that’s where you may have to look at something that’s having either slower ramp rate at certain parts, for example, if you froze down to -10 or -15 to kind of encourage nucleation to start, and then once you’re down there, either holding it at -15 or -20 for a couple hours, so you’re freezing at a slower, more steady rate and then lowering to your final temperature, or lowering down to maybe -15°, and then from there on, continuing at maybe a quarter of a degree per minute for the next 10° or so. So you’re again trying to keep a steadier temperature difference between your product and your shelf for a steady rate of freezing, which you can do by, you know, changing up that rate of freezing, but, I mean, it is something that’s definitely not as easy without the ability of controlled nucleation, which is something in the industry right now, where, you know, on the production side, they don’t have controlled nucleation available. So it’s tough to try to develop a cycle using controlled nucleation and then on the other side, you know, a lot of people are looking at investing in controlled nucleation on the production side if they haven’t developed cycles with a controlled nucleation on the R&D side. So it is about a little bit of a chicken and egg there. I would definitely say I recommend looking into controlled nucleation and if that is something that could be feasible and beneficial for your product and your process, but there are things you can do without it that may improve, to different degrees, the ice structure within your product, and improve—help optimize—the freezing side of that freeze drying cycle.
CZ: Okay, here’s the second question from Jody. Jody’s asking, it looks like a lot of freezing and primary drying optimization is based on AccuFlux and AutoDry. How could we transfer an optimized cycle using those features if they are not present on our production freeze dryer?
SH: That’s a good question. Maybe something I made plans to go over in the presentation originally. So when you’re looking at the features such as AccuFlux or AutoDry—which are automatically controlling the shelf temperature based on the product temperature, and for AutoDry in primary drying, or controlling shelf temperature based on your heat flux setting for Accuflux during freezing—those programs are automatically going to be controlling that shelf temperature, but they may not do so identically between different runs, you know, based on a lot of the variance of the way they run. We don’t expect that AutoDry and AccuFlux will give you identical profiles each time you run them. So really, what they’re meant for is as a guide for then converting into an optimized recipe. So for the example of AccuFlux, what we typically see in an AccuFlux profile is, that after controlled nucleation, depending on where your heat flux setting is, either AccuFlux will maintain the shelf temperature at the same temperature it was at, or it may reduce the shelf temperature a little bit off the start, right after nucleation, to push that heat flux down to your set level. But then once that heat flux is at the level that you set it at, it typically does not make rapid shelf temperature adjustments, because when your product is freezing, it’s staying very near the same freezing temperature—close to 0°C or -1, depending on your concentration. And so the shelf temperature then will, for the large part, be pretty steady to maintain a constant temperature difference between your product and your shelf as it is then done, or as it then nears the end of freezing and the product temperature starts to drop, AccuFlux will slowly ramp your shelf temperature down to maintain a steady temperature differential. So one example of how you may transfer that into a recipe is after nucleation, either maintaining your same supercooling shelf temperature at -5 for example, and holding it there for an hour or two, and then ramping down the shelf further. Or after controlled nucleation, taking a shelf temperature and ramping it down, you know, maybe as slowly as a tenth of a degree per minute for an hour or two to try to maintain a relatively steady rate of freezing and a relatively steady temperature difference between where your product is and what your shelf temperature is.
Similarly on the side of AutoDry, what AutoDry will tell you is essentially how high you can bring your shelf early on in the set point at the highest temperature that AutoDry reaches, and what a safe temperature to finish drying is. So we recommend two different ways of transferring an AutoDry cycle. A conservative method would be to set your shelf temperature for primary drying at whatever the final shelf temperature for AutoDry was. So, for our example, that would be -14 or instead of our conservative baseline temperature of -25, maybe we’ll just take it straight up to -14 and hold it there for the cycle because we know that we’ll be safe. Or a little bit, slightly more aggressive way to transfer an AutoDry cycle would be to ramp your shelf up initially somewhere near the max temperature of auto drying. So in our example, that was -3 maybe to transfer this to a recipe that we know we’d consistently run safely, you know. For many runs to come, maybe we’re not going to ramp it up to exactly -3, but maybe we’ll go close, you know, ramp it to -5ׄ°C for an hour or two, and then bring it back down to that safe ending temperature of -14, or again, with a more safe offset, maybe -15, -16 for the remainder of the cycle. So you can try to capture that first large bump, that large bump in temperature early on in the cycle, and then lower it down later. So that can be converted in a variety of ways, ranging from relatively conservative to more aggressive for AutoDry. But in either case, with AccuFlux or AutoDry, the objective is to use the profiles from those programs to give you an idea of how to set your recipe that will produce a similar product profile and set up a similar optimized cycle.
CZ: Okay. Our next question is from Lucy. What is the difference between using the shelf surface or shelf inlet temperature for calculating and comparing kVs?
SH: The difference between using kV based on a shelf surface or your shelf inlet temperature is really just the difference of the point of reference you use, and, you know, I can’t say definitively say which one is better than the other. For our example, in the test we did here, in the transfer example we did here, we used the shelf surface temperature because part of the heat flux sensor that we’re using has a thermocouple on it that does give us a reading of the temperature at the actual shelf surface. So whichever one you use, it is not really critical. The important thing is that it’s consistent between whatever kVs you’re comparing. And you may find, you know, depending on the application, which one may be more convenient than the other one, and that’s primarily the reason for picking one over the others—is convenience. So again, for example, we have the shelf surface temperature information as part of that heat flux sensor integrated in the surface of the sensor. It’s flat and not intrusive, so it’s easy to get to so we can use the shelf surface temperature. In some cases it may not be as easy to attain the shelf surface’s temperature for something like a production unit, where you can try, you know, taping a thermocouple with a little bit of a thermal paste to the shelf. But, you know, it’s a different question of whether that’s accurate compared to what the surface temperature will be beneath the sublimating vial. So in cases like that, where it’s difficult to get the shelf surface temperature, it may be easier to use the shelf inlet temperature. And when it comes down to it for transfer, again, like I said, it’s not critical which one you use as long as they’re consistent. So when you’re using a shelf inlet temperature—when you’re comparing shelf inlet temperatures, or kVs based on shelf inlet temperatures between two units, looking at the ratios of those kVs and the shelf inlet temperature of your starting unit—your source unit—will then give you, or allow you to calculate, the ideal shelf inlet temperature on your target unit, which is what control the shelf temperature is based on—it’s based on inlet temperature, so it’s direct and it’s easy to figure out what to put in your recipe and your target unit if you know what your target shelf inlet temperature is. If you’re using the shelf surface temperature and you’re comparing kVs based on the shelf surface temperature between two units, and you’re looking at the shelf surface temperature of your source unit and the ratio of those kVs, that’s going to allow you to calculate what the ideal shelf surface temperature on your target unit is going to be. Now that you know the ideal shelf surface temperature, you know what that difference is between the inlet and the surface temperature to then figure out what to put in your recipe to control the shelf temperature around, based on. So it’s an extra little step, and that was that ∆T term in the equation we looked at because we were using shelf surface temperatures. Had we been using the inlet temperatures, that term would have dropped out because it’s kind of built into the kV based on the shelf inlet temperature. So in that case, it may be more convenient, in general, to use the kV based on the inlet temperature. Again, for the transfer example we had here, we had a heat flux sensor in both our source and target unit, so it was easy to use our surface temperature, but maybe not the case for, or likely not the case, for most applications, in which case we would probably recommend using the inlet temperature for your kV as your kV basis.
CZ: Okay. I think that’ll do it for the questions today. I’d like to thank everyone again for attending today’s webinar, and we look forward to seeing you again at a future McCrone Group webinar. Thank you.