Ice Fog as a Means to Induce Uniform Ice Nucleation During Lyophilization (Peer Reviewed)

ABSTRACT

Lyophilization or freeze drying is an important downstream process for stabilizing pharmaceutical compounds. The control and repeatability of lyophilization cycles are crucial for achieving consistently high product quality. Although the obvious parameters of shelf temperature and chamber pressure may be well controlled, the lack of control of the ice nucleation temperature (the temperature at which the product freezes) can adversely affect product uniformity and lead to suboptimal freeze-drying cycles. This study describes a novel means to control ice nucleation using a sterile cryogenic ice fog that is applicable to laboratory-, pilot-, and production-scale lyophilizers. Test results demonstrate the scalability and robustness of this technique.

Lyophilization or freeze drying is an important downstream process for stabilizing pharmaceutical compounds. It involves removing water and solvents from a product by sublimation and desorption to levels that will not support biological or chemical reaction. It is an excellent method to extend the shelf life of sensitive compounds for storage and transportation without subjecting them to detrimental high temperatures, and the only method available for a majority of biological compounds. Consequently, lyophilization continues to be indispensible to the pharmaceutical industry, despite its high cost and complexity.

Lyophilization consists of two major steps: freezing solutions, and drying the frozen solid under vacuum through sublimation and desorption. The drying step is divided into two phases: primary drying (i.e., ice sublimation) and secondary drying (i.e., liquid desorption). A successful lyophilization cycle can be defined by dried product that is visually and functionally acceptable with a short reconstitution time, potent active ingredients, and increased shelf life. The control and repeatability of the cycle are crucial for achieving consistently good product quality. Demand for lyophilization technology is growing because of the high value of the drugs being lyophilized as well as FDA initiatives such as quality by design (QbD) and process analytical technolgoy (PAT). Consequently, the industry has been quick to develop and adopt technologies that facilitate improved control of key process parameters. Controlling ice nucleation during the freezing cycle of lyophilization is one such tactic that is currently under investigation as a means to achieve more robust and scalable lyophilization cycles.

The importance of ice nucleation temperature

The onset of freezing, or ice nucleation, is one of the most important steps in the lyophilization cycle. For nonaseptic systems a particle or impurity often serves as the nucleation point that allows ice crystals to grow and the product to freeze. However, in aseptic systems of high purity the product sometimes cools below its freezing temperature without ice crystal formation because no particulates are available for ice nucleation, a process known as supercooling. Substances that cool below the freezing temperature without becoming solid are referred to as supercooled. The degree of supercooling determines the ice crystal structure, which in turn characterizes product resistance to water vapor flow during the drying cycle. Increased supercooling has been shown to form smaller, more numerous ice crystals, thus resulting in higher product resistance and increased drying times. Studies have shown a 1–3% increase in primary drying time for every 1 °C decrease in ice nucleation temperature (1, 2). Supercooling of vials during freezing can thus increase cycle times and operating costs.

Lack of uniformity in ice nucleation temperature caused by vial supercooling can lead to vial-to-vial variability in ice crystal structure. Vials that freeze at high temperatures dry faster than those that freeze at low temperatures, making it difficult to have a drying cycle that is optimal for all vials. This variability causes problems such as vial breakage and melt-back, and decreases overall yield and product uniformity.

In addition, variability in ice nucleation increases the uncertainty in scaling up a cycle from laboratory (nonaseptic) to production scale (aseptic). A cycle optimized at lab scale may have entirely different drying time requirements at production scale due to the higher degree of supercooling expected in particulate-free, production-grade environments. Variability in ice nucleation is compounded by vial-to-vial variations in drying behavior due to variable ice structure.

Methods to address issues related to nonuniform ice nucleation

Although ice nucleation is an important parameter for achieving robust cycles, there have been very few attempts to achieve it at commercial scale until recently. The standard practice has been to use an annealing cycle, which involves raising the product temperature after freezing to a temperature above glass transition, and then holding. This method results in the formation of larger ice crystals at the expense of smaller ones, and helps minimize the variability in drying behavior. However, annealing may not be well tolerated by protein systems that are susceptible to denaturation. In addition, the benefits of shorter drying times may be offset by the additional time required for the annealing cycle. Lastly, annealing fails to address the root cause of variable ice structure, which is the lack of a uniform ice nucleation temperature, and can only help to repair the damage already caused.

Other methods that have been tried at laboratory scale include using nucleating agents such as silver iodide and bacteria, ultrasonic vibration of the product, etched vials, sudden depressurization, and ice fog. This study will focus on the last method, ice fog, and show its successful transition from a laboratory concept to a commercially viable technique.

Ice fog as a means to induce uniform vial-to-vial ice nucleation

As discussed above, one approach for reducing supercooling and controlling ice-nucleation temperature is to introduce nucleating particles into the supercooled solution. A particularly advantageous nucleating particle is microscopic ice (i.e., frozen water) crystals in the form of a fog introduced into the freezing chamber (3). The concept of temperature-controlled ice nucleation was suggested by T.W. Rowe in 1990 (4). A cryogenically created fog containing microscopic ice crystals is introduced into the lyophilization chamber after the vials have reached the temperature at which nucleation is desired. The ice crystals subsequently make their way into the vials and induce nucleation inside the vial. Although this technique has found success on a laboratory scale, it has proven difficult to scale up to commercial lyophilizers. The difficulty is not only forming the ice fog and ensuring it is sterile, but also uniformly distributing the ice fog rapidly throughout the freezing chamber so that all vials are properly seeded with nucleating ice particles.

This article will describe a means to produce and distribute an aseptic ice fog that nucleates all vials in a short time. This work has resulted in a novel means to produce and distribute a sterile ice fog that is applicable to laboratory-, pilot-, and production-scale lyophilizers. This scalable cryogenic ice fog technology could provide a much-needed degree of control during lyophilization and thus facilitate application of QbD principles in this crucial downstream operation.

METHODS

Figure 1 is a schematic illustration of a typical lyophilization system employing the scalable cryogenic ice-fog technique. Creating a uniform dispersion of ice fog, distributing it into the freezing chamber and seeding vials with ice crystals for nucleation are achieved by a patent-pending technique involving contact between liquid nitrogen and water in a mixing device such as an ejector, outside the lyophilization chamber (see Figure 1). The ejector circuit is composed of a port for introducing ice fog into the freezing chamber and another port for recycling fog out of the chamber.

Figure 1: Illustration of a typical lyophilization system employing the scalable cryogenic ice-fog technique. (ALL FIGURES ARE COURTESY OF THE AUTHORS)

Ice-fog introduction followed the two-step approach shown in Figure 2. The vials containing the product to be freeze dried were placed on the cold plates inside the freezing chamber. In the initial phase of the freezing process, the vials were cooled to a suitable temperature at or below their freezing point. When the suitable vial temperature was achieved, a cryogenic ice fog was introduced into the chamber for about 30–50 s. Detection of ice nucleation in the vials was assessed by a combination of direct observation and temperature measurements on the outside of select vials. The metal door of the lyophilizer was replaced with a Plexiglas construction to facilitate visual observation and video recording.

Figure 2: Illustration of the two-step approach for ice-fog introduction.

In control experiments, the normal freezing cycle was run with no ice fog introduction. The goal was to determine the shelf temperature at which the first vials nucleate and freeze. This shelf temperature in subsequent trials helped determine the trigger temperature (tg) that indicated when the ice fog should be introduced into the chamber. The experiment also showed the extent of subcooling and vial-to-vial variability in freezing temperature by recording the range of temperature and time over which all vials nucleated.

Two sets of tests were performed using two lyophilizers. The first set was performed in a MINIFAST 1.0 (IMA Life) with 1.1 m² of shelf area and represented a laboratory-scale lyophilizer. The second set was performed in a LYOMAX 2.5 (IMA Life), with 2.5 m² of shelf area and represented a pilot or commercial-scale lyophilizer. Prefilled sterile vials were obtained for the testing, with between 10–20 vials arranged to be visible from the front of the chamber. Some vials were also strategically placed inside the lyophilizer on areas of the shelves where distribution of ice fog was expected to be most challenging. One of these vials was designated as the trigger vial (see Figure 3, Note 1). The temperature of this vial was monitored to determine when ice fog should be introduced into the chamber.

Figure 3: Temperature measurements obtained in control experiments as a function of time.

Of the total vials, nine were instrumented using K-type thermocouples. Vials without temperature probes were observed visually (see Figure 3, Note 4). All instrumented vials, with the exception of the trigger vial, had thermal probes mounted on the outside of the vials and touching the vial wall. The trigger vial contained the temperature probe inside the solution, but not touching the walls or bottom of the vial. Thus, it gave a truer indication of the solution temperature compared with the other instrumented vials where the temperature probe was mounted on the outside. However, it was also the vial most likely to freeze first because the probe itself served as a point of nucleation. tg was set at the temperature at which the trigger vial froze.

Two populations of vials were used in the same test. One population was filled with pure water only (see Figure 3, Note 2), and the other was filled with a solution comprising 5% glycine and 1% NaCl (see Figure 3, Note 3). All solutions were filter sterilized through standard 0.22-µm filters before use. A modular cleanroom was constructed around the front side of each lyophilizer to replicate the particulate-free condition of production-grade environments. A laser counter was used to measure the particle concentration inside the clean room, and it measured particulate (i.e., > 5 µm) impurity concentration to be under 15 particles/ft³.

RESULTS AND DISCUSSION

Creating and uniformly distributing cryogenic ice fog

A key challenge for the commercial implementation of the ice-fog technique has been the creation of an ice fog that is sufficiently dense and that can be efficiently distributed to reach all vials in a large-scale lyophilizer. Because of this low density, not enough fog was available for all vials. Also absent in previous tests was an efficient system to distribute the ice fog within the freezing chamber and drive it into the vials. The system used in the present study produced a very dense fog and also distributed the fog throughout the freezing chamber within a short time (i.e., less than a minute). It is also possible, through this design, to control the density of the ice fog.

Fog creation and distribution were aided by the ejector assembly. The ejector serves two purposes. First, it provides an extremely efficient means for quickly forming the ice fog. Second, the suitably sized ejector provides enough pumping capacity to circulate the ice fog throughout the freezing chamber rapidly. It is a significant advantage that the ejector can accomplish both of these functions without introducing any moving parts or other complicated mechanisms that would be difficult to steam or otherwise sterilize.

Achieving ice nucleation in all vials at desired temperature

Figures 3 and 4 show the temperature measurements obtained in control and ice fog experiments, respectively, as a function of time. Ice nucleation was indicated at the point when the temperature of a vial increased sharply. This result is due to release of the latent heat of fusion of the solution upon freezing.

For the control experiment, the first vial nucleated at a temperature of around –9 °C and the last vial nucleated at around –18 °C (see Figure 3). About 20 minutes separated these two occurrences, and the remainder of the vials nucleated at various times inbetween. This variation in ice nucleation time could increase in production-grade environments, where solutions may be supercooled further due to the absence of any particulates or impurities in the atmosphere.

Based on the data from control experiments, a trigger vial temperature of –6 °C was selected as tg. The choice of this tg was conservative so that all vials were cooled below their freezing point. This choice was to ensure that absence of freezing was a result of supercooling only, and not because a vial was at a temperature above the freezing point. Sometimes uneven vial temperatures may occur in laboratory-scale lyophilizers because of nonuniform shelf cooling. In addition, the temperature probes, except the trigger vial probe, measured the outside vial temperatures which might not reflect the solution temperature inside the vial at all times. If all vials are cooled below their freezing point, ice fog can be introduced at a much higher temperature below 0 °C.

Shelves were allowed to cool at a ramp rate of 0.5 °C/min, and the temperature of the trigger vial was constantly monitored. When trigger vial temperature hit –6 °C, the ice fog was introduced. As seen in Figure 4, all vials nucleated at the same instant following the introduction of ice fog. Both vial populations, pure water and glycine solution, nucleated at about the same time on ice-fog introduction, indicating the general applicability of this technique for all supercooled solutions.

Figure 4: Temperature measurements obtained in ice-fog experiments as a function of time.

From a regulatory standpoint, creation of the ice fog at production scale does not introduce anything fundamentally new to the system. The ice fog is produced inside the ejector using steam and sterile-filtered nitrogen gas, both of which are already used in lyophilizers today (e.g., steam for sterilization and nitrogen for inerting or backfilling). All components downstream of the sterile nitrogen gas filter and up to the output of the ejector that releases the ice fog into the lyophilizing chamber have been designed to be sterilized in place. Hence, all the surfaces the ice fog touches before being introduced into the lyophilizer are sterile. All surfaces within the lyophilizer itself, including the vials, are sterilized, and the ice fog does not touch anything that is nonsterile, even after being introduced into the freezing chamber. In summary, the introduction of a sterile ice fog is no different from the introduction of any inert gas, such as nitrogen, that is used today for backfilling vials. No additional sterility concerns should arise regarding the surfaces the ice fog touches inside the lyophilizer.

Introducing water in the form of ice crystals into a finished formulation may raise concerns initially. However preliminary tests have shown that ice-fog derived water is a small fraction of the total water already present in the formulation, and comparable with the prevalent chamber moisture content that formulations routinely encounter when loaded into lyophilizers.

Ice nucleation in all vials was further confirmed visually and through video recording. Figure 5 depicts a 7-s video as a sequence of still frames separated by 0.4 s in real time. It shows the lyophilizer being filmed from outside the plexiglass door during the introduction of the ice fog. The first image in the sequence shows the chamber before the introduction of the ice fog, and the last image shows it 7 s after the introduction of the ice fog. The images clearly show a dense ice fog distributed throughout the chamber within this time.

Figure 5: Sequence of still frames from a 7-second video in increasing order of time from left to right. The first image shows the chamber before the introduction of the ice fog. The last image shows the chamber 7 seconds after the introduction of the ice fog.

Ice nucleation inside the vials can be visualized in Figure 6, which shows a 4-s video as a sequence of still frames separated by 0.3 s in real time. It shows the close-up of three consecutive vials placed in the center of the middle shelf of the lyophilizer, where ice fog reach is expected to be the most challenging. The first image shows the close-up of one vial just as it begins to nucleate after the introduction of the ice fog. Within 4 s, vials adjacent to it also nucleate and at the end of 4 s, all three vials have completely nucleated. On a macro scale, this phenomenon is seen in all vials inside the chamber, and all vials nucleate within 4–10 s following the introduction of ice fog. This result is a significant improvement over the 20-min vial-to-vial nucleation variability seen in the absence of ice fog.

Figure 6: Sequence of still frames from a 4-second video in increasing order of time from left to right. The first image shows three consecutive vials placed in the center of the middle shelf in the lyophilizer before introduction of the ice fog. The last image shows the same vials 4 seconds after the introduction of the ice fog.

Scale-up considerations and potential regulatory concerns

The scalability of the technique has been verified by replicating it on a lab-scale (MINIFAST) and a pilot-scale (LYOMAX) lyophilizer. It is expected to be easily scalable to larger sizes. The water-vapor source for ice-fog generation can be chosen based on ease of use and infrastructure availability. For instance, in nonindustrial, nonaseptic laboratories, a humidified gas stream may be the preferred source, whereas on the aseptic production floor, steam would be the preferred fluid.

CONCLUSION

Ice nucleation during vial freezing in lyophilization is an important process parameter that needs to be controlled. The scalable cryogenic ice-fog technology can be used in laboratory-, pilot-, and production-scale lyophilizers to induce uniform ice nucleation and eliminate vial-to-vial variability. Eliminating variability, in turn, can help mitigate a host of related issues and lead to improved process and product quality.

Prerona Chakravarty, PhD*, is a project manager, and Ron Lee, PhD, is a research fellow, both in Pharmacueticals, Fine and Specialty Chemicals in Linde Gases Division, Murray Hill, NJ. Frank DeMarco is freeze drying development manger, and Ernesto Renzi is president of sales, both at IMA LIFE North America, Tonawanda, NY. *To whom correspondence should be addressed, prerona.chakravarty@linde.com.

PEER REVIEWED

Article submitted: Aug. 16, 2011. Article accepted: Oct. 18, 2011.

REFERENCES

1. M.L. Roy and M.J. Pikal, J. Parenter. Sci. Technol. 43, 60–66 (1989).

2. J.A. Searles, J. F. Carpenter, and T.W. Randolph, J. Pharm. Sci. 90, 860–871 (2001).

3. S. Rambhatla et al., AAPS PharmSciTech 5 (4), 54–62 (2004).

4. T. D. Rowe, presentation at International Symposium on Biological Product Freeze-Drying and Formulation (Geneva, 1990).