Lyophilizer Shelf Temperature Mapping: Part II: Study Design

INTRODUCTION

The US FDA has provided several guidelines on lyophilizer qualification, however, there is little detailed guidance in the literature or regulatory documents regarding how to perform a shelf temperature mapping study1. Part I of this series of publications identified appropriate temperature measurement devices and chamber environment conditions for executing a shelf temperature mapping study.

In this article, additional variables were evaluated to determine how to best conduct an appropriate study and provide direction to answering the following questions:

• How many locations are needed to accurately map the shelf?
• How should the mapping locations be selected?
• What temperatures should be evaluated?
• How long should the measurements be taken?

Shelf Mapping

When designing studies for shelf temperature mapping, it is important to consider where the temperature probes should be placed to obtain an adequate temperature map of the shelf surface. It is extremely difficult and unreasonable to attempt to map the entire surface of the shelf. US FDA guidance indicates that the four corners and center of shelf are typically measured2. Because this is not an essential requirement, a sound scientific rationale based upon empirical data needs to be developed to justify the locations where the shelf surface temperature is being measured. Studies were undertaken to evaluate the edge effect on the surface temperature as well as the temperature across a shelf. For this purpose, a 24-inch-by-36-inch (610-mm-by-914-mm) shelf was mapped in two different configurations with up to 20 temperature probes, such that a significant portion of the shelf surface could be mapped. Temperature probes are a wireless device with a 1⅜-inch (35-mm) diameter stainless-steel base and an RTD secured into it at the center of the base. Capabilities of such temperature probes were evaluated and their effectiveness assessed in part 1 of this series. The temperature probes were placed on a single, middle shelf according to the diagrams in Figures 1 and 4. Measurements were made at -50°C and 50°C with the chamber pressure set at a target pressure of 200 µm Hg.

Analysis of the temperature data should only occur once the shelf and the sensors have reached a steady state condition. Temperature data at the beginning of the test interval at a given target temperature is inconsistent and unreliable. Jennings et al. suggests that data analysis should occur after every individual measurement does not vary by more than ±1°C over a period of 30 minutes6. The test should then be conducted for a period of time to ensure there is not drift to the data over time. A 60-minute measurement time for each temperature, after equilibration is achieved, is sufficient to collect data for analysis. It is recommended to conduct an initial test to determine the time required to reach steady state at each temperature evaluated and apply an equilibration hold time in addition to the 60-minute measurement time in the test protocol.

The lyophilizer used in these studies is comprised of four shelves, each providing 6 ft2 (0.56 m2) of shelf surface area each, for a total of 24 ft2 (2.23 m2). The shelves are designed with three internal stay bars to direct the heat transfer fluid in a serpentine path through the shelf. The stay bars and edges of the shelf are approximately ¾-inch (19 mm) wide and are placed every 6 inches (152 mm) across the shelf. Two shelf maps were evaluated to cover potential influences of the shelf design and construction (see Figure 1). The maps reflect the locations on shelf 2 of the four-shelf internal condenser lyophilizer. The corners and edges of the shelf were mapped to evaluate the edge effect as compared to various locations throughout the interior shelf surface.

The test was conducted using both temperature maps and the results were evaluated over a 1-hour measurement interval with a target shelf inlet temperature at the high and low temperature range of typical operation, -50°C, 0°C, and 50°C, respectively (Figures 2 and 3).

With the target setpoint at a low of -50°C, each corner at the very edge of the shelf showed the greatest difference relative to the target setpoint, with the warmest location at the front right corner, 3.6°C higher than the setpoint. The temperatures along the sides of the shelf were indistinguishable from the temperature at the center of the shelf. However, the front and back edge and particularly the corners of the shelf were as much as 2.6°C different from the center measured temperature. There was no meaningful difference in all the center locations, including those sensors over the stay bars. The temperature variation at all other locations was within 2°C of the target setpoint and the interior locations well within 1°C.

Similar to the results of the shelf temperature mapping at -50 °C, the right front corner of the shelf showed the greatest variation from the target setpoint at 50°C with as much as 3.4°C colder than the setpoint. All the corners and the front and rear center edge were also the locations with the greatest difference as compared to the target setpoint of 50°C. The temperature variation at all other locations was within 2°C.

Conversely, at the mid-point of the range, 0°C, the results of the shelf temp-erature mapping show little deviation from the setpoint, from -0.4°C to 0.3°C. The front center of the shelf is the warmest location while the right edge of the shelf is the coldest. The temperature variation was within 1°C. It is apparent that as the shelf temperature gets closer to the outside environment, the edge effect is lessened dramatically.

Figure 1B reflects locations selected to be over top of the stay bars and in the middle of the heat transfer fluid flow path as well as at each turn in the serpentine path of the heat transfer fluid. The objective of measuring such locations in this pattern was to assess any temperature variation across the shelf as well as across the front and rear locations. Evaluation of any impact of the stay bars on heat transfer to the shelf surface is also assessed.

Figure 3 reflects the shelf surface temperature measurements based on the locations in Figure 1B over a 60-minute interval. The data indicate that the temperatures across the center of the shelf were indistinguishable from each other regardless of the shelf setpoint. The front and back edge and particularly the corners of the shelf were between as much as 3.0°C from the target setpoint of -50°C and 2.6°C from the target setpoint of 50°C, while at 0°C the corners were only up to 0.1°C from the setpoint. As in the previous studies, the extreme edges were significantly different than the other shelf locations. Temperatures at a diagonal location relative to those of the previous study with the measurement adjacent to the extreme corner revealed a less significant difference. The data again shows that the stay bars do not contribute significantly to temperature variability on the shelf surface.

The initial study was intended to evaluate the temperature profile in an array of locations. The follow up study was to measure both the shelf surface at the corners and locations at close proximity across the shelf. Results from both mapping studies suggest the corners, front edges, and back edges of the shelf are different from the center. This temperature difference is likely influenced by radiative heat transfer effects due to the surrounding environment and should be considered when analyzing the study results. The edge effect is more pronounced at extreme temperatures and along the front edge of the shelf than the back edge. At the extreme temperatures of -50°C and 50°C, the edge effect can introduce variability of 1°C to 3°C along distances of less than or equal to 3 inches (76 mm). This is to be expected as the door typically has the least amount of insulation of any part of the chamber and allows more heat flux influence on the devices used for shelf surface temperature measurements. It is interesting to note that the temperature differences at the corners does not occur with a shelf temperature of 0°C.

With measurements across the middle of the shelf (Figure 3), where the temperature probes were placed at equal distances from each other with approximately 1.9 inches (47.6 mm) between each probe, there was a nominal temperature difference. The differences were within a few tenths of a degree, even at the side edges and at the extreme temperatures. All the temperatures were well within the calibration tolerance of 0.5°C.

With an understanding of the shelf mapping studies intent, a justification for an effective mapping configuration can be established. The design needs to be based on knowledge of the shelf construction, heat transfer fluid flow, confidence in the accuracy, and precisions of the shelf surface temperature measurements and supported with a scientifically sound rationale.

Based upon an understanding of the shelf design, sensors placed over the middle of the flow path at the inlet and outlet of each shelf, which, in the lyophilizer used in these studies, are the back corners about 3 inches (76 mm) in from the edge. This placement would measure the shelf surface temperature where the heat transfer fluid first enters and where the fluid exits a shelf. The measurement at the inlet would be expected to represent temperatures nearest to where the control and monitoring temperature probe is located, and therefore closest to the target setpoint. The measurements at the shelf outlet would be expected to reflect any change in temperature as the heat transfer fluid flows through a shelf. For the configuration in the shelf construction of the lyophilizer studied, two additional sensors were placed near the front corners of each shelf as these locations measure how the shelf temperature would change with the heat transfer fluid entering and traveling from the inlet to the front of the shelf and the fluid leaving the shelf. This would also represent where heat transfer fluid would encounter significant turbulent fluid flow. Since the extreme edge of the shelf is highly affected by environmental radiative heat flux and does not represent the performance of the heat transfer system or an assessment of any shelf temperature mapping results, it is recommended to keep the temperature probes between 1 inch (25 mm) and 3 inches (76 mm) from each corner of the shelf. A fifth sensor could be placed at the geometric center of each shelf as a measurement to reflect temperatures in between the other four measured locations and provide sufficient measurements to calculate a more meaningful range and average.

Test Temperatures

Various temperature target setpoints are employed in the shelf temperature mapping tests by different freeze dryer manufacturers and reflected by those referenced in the literature. Huang indicates the temperature range should exceed the target setpoints used for routine lyophilization4. Jennings et al. suggests testing at the minimum and maximum achievable shelf surface temperatures as well as at temperatures of -40°C, -20°C, 0°C, and 20°C6. Several freeze dryer manufacturers specify a minimum temperature of -40°C for their specification. Fischer recommends monitoring at the “worst-case” temperatures of all recipes, in which those Fischer referenced were -45°C and 60°C7.

Given the variability in the ranges used, studies were conducted using the lyophilizer described earlier to determine if the shelf temperature setpoint changes the shelf surface temperature uniformity observed. Temperature probes found to be most effective in part 1 of this series were used to measure the shelf surface temperature while at the shelf temperature setpoint of -53°C, -50°C, -47°C, -44°C, 0°C, and 50°C. For these studies, the temperature range was chosen to encompass the normal operating range of the freeze dryer. The temperature probes were placed approximately 3 inches (76 mm) inside each corner and in the center of the shelf for a total of five temperature probes per shelf; a total of 20 temperature probes were used to be able to evaluate four shelves. After equilibration at each shelf inlet temperature, the shelf temperature was monitored for 60 minutes, and the average shelf temperature at each location was calculated. Equilibration was established as no more than 0.2°C change over a 60 minute period. The average shelf temperature over a 60 minute period was then compared with the average shelf inlet temperature over the same 60 minute period.

Figures 4 is a chart of the difference between the maximum shelf temperature for each of the four shelves relative to the average temperature measured at the shelf inlet. The data are graphed sequentially from the coldest target setpoint to the warmest.

The temperature range at the target setpoints for which the study was conducted is an important factor in the experimental error for the claimed shelf surface temperature uniformity. The data from these studies shows there is an increase in the difference between the surface temperature measurement and the shelf inlet when the target temperature is set at the extremes of the operating range of the dryer, in this case -53°C and 50°C. This data demonstrates evaluating the extremes of the temperature range of the shelf surface temperature, provides a high degree of confidence in the shelf temperature uniformity when implementing routine operating during processing. The study conducted at 0°C reflects the shelf temperature uniformity at more typical operating temperatures. The temperature selected for the studies may be to envelop the expected lyophilizer operating range or challenge the vendors claimed performance capabilities. It is important to recognize the resulting temperature measurements may depend on the lyophilizer size and capacity, temperature control strategy, and range of temperatures used during routine processing. It is reasonable to, at a minimum, challenge the equipment at the warmest and coldest temperatures intended for processing, as well as the midpoint of the temperature range. If the claimed shelf uniformity range is ± 3°C from the target setpoint it would be of value to challenge the system 3°C beyond the specified operating range. This generates data to evaluate the full equipment operating capability.

Specifications

When performing a shelf temperature mapping study, the goal is to ensure a uniform temperature distribution across the lyophilizer shelves to demonstrate the heat transfer system performance as a “baseline” under no-load conditions. This is valuable data for future reference as part of a change control strategy. It is therefore reasonable to question what is an acceptable shelf temperature uniformity. A search of the literature reveals there is no consensus, or even guidance to a suitable range for a shelf surface temperature variation, although opinions have been expressed in the literature.

Huang suggested that a limit may be 1–2°C on each shelf and between the shelves and that cold and hot spots should beidentified4. Jennings et al. suggests that the temperature distribution should be verified6. Rambhatla et al. suggest that the shelf temperature distribution be used to define “hot” and “cold” spots on the shelf and the difference between the shelf inlet and these spots be used as a scale-up tool for defining changes to the process parameters. However, the authors do not present a justification or rationale, nor any correlation to a product impact8.

It would be reasonable to refer to the engineering experience and knowledge of those most well versed in the equipment and the process for directives in a suitable shelf surface temperature variation. Freeze Dryer manufactures set specifications for the shelf temperature uniformity as factory acceptance criteria. For example, several manufacturers have different criteria. There is no published scientific rationale or technical justification for the chosen criteria.

The temperature probe data collected at the previously defined 20 locations throughout the dryer in four separate studies at each target shelf temperatures was analyzed for variability from run to run. The location having the most extreme temperature relative to the average of all the measurements (extreme) and the location closest to the average of all the measurements (representative) were identified in the 24 ft2 (2.2 m2) lyophilizer.

The results of the four studies are summarized in Table 1. The standard deviation across the four studies at each location ranged from 0.1°C to 0.5°C. There was little consistency in the location of the extreme and most representative results. At -50°C, shelf 4 as a whole appeared to consistently be warmer than the other shelves (Figure 5A) with the center location being the most extreme location in the whole dryer in three of four studies; however, at 50°C, no similar trend was observed. This suggests that identifying “hot” and “cold” spots would not be possible in this lyophilizer, and possibly any lyophilizer, as the variation in the data at different shelf temperatures masks any variability in the actual temperature distribution across the shelf (Figure 5). As well, the impact and significance of a “hot” spot and “cold” spot across the temperature range of -50°C to 0°C to 50°C could easily be different.

When setting the acceptance criteria it is also important to consider how the system is being controlled. Figure 6 shows the difference between the shelf inlet temperature and the target setpoint, in the 24 ft2 (2.2 m2) pilot-scale lyophilizer with an internal condenser, at three different setpoints spanning a typical operation range. The variation from the setpoint is different at each target temperature; thus, the shelf surface temperature will also be expected to vary.

When performing a shelf temperature mapping study, the goal is to ensure a uniform temperature distribution across the lyophilizer shelves to demonstrate the heat transfer system performance as a “baseline” under no-load conditions.

When considering the acceptance criteria for a shelf temperature mapping study, it is important to understand the behavior of the lyophilizer throughout the temperature range being tested as well as any configuration details that may influence the measurements such as internal condenser or sensors near the nitrogen inlet port(s) used for pressure control. As well, the variation of the shelf inlet temperature may be different depending on the target temperature, control strategy, and lyophilizer capabilities. The data collected in the 24-ft2-(2.2-m2-)unit shows the range of the measurement from one run to the next is more significant than any variability inherent in the locations (see Table 1). Analysis of the data and evaluation of the results are discussed further in part 3 of this presentation.

Any acceptance criteria established should account for the variability in the measurement plus any variability inherent in the calibration of the temperature sensor. Typical tolerance for calibration of temperature measuring devices may range from ±0.5°C to ±1.0°C. Therefore Guard Banding based on the tolerance of the calibration of the measuring devices should be considered when setting the acceptance criteria for the shelf temperature variability9, 10. The data indicate that the variability in the measurements are anywhere from ±0.1°C to ±0.5°C. Therefore, it would be unreasonable to consistently expect variability of less than ±1.0°C: Based upon the cumulative effect of measurement and calibration error, the variability could be as large as ±1.5°C.

CONCLUSION

The studies conducted demonstrate the factors that may influence the results when designing a shelf temperature mapping study. The temperature probe used should be selected to measure the shelf surface temperature with sufficient accuracy and precision. A rationale should be established for the locations selected for mapping the shelf because it is unreasonable to map the entire shelf surface. Every shelf should be mapped and if multiple studies are required to map all the shelves, then one shelf should be included in every study to allow comparison of each shelf to a common shelf surface reference to ensure consistency across all the individual studies.

When selecting the target temperature for evaluation, the typical operating temperature range should be understood and considered. The study should encompass both extremes of the operating range, with at least one temperature in the center of the range. When designing the study, steps should be taken to mitigate the effect of the chamber environment on the shelf surface temperature measuring probe so that the study is truly evaluating the uniformity of the shelf and performance of the heat transfer system.

The acceptance criteria and data analysis should consider the inherent variability of the measurement and calibration tolerance, as well as any other factors that might influence the accuracy and precision of the measurements and analysis of the results. Shelf surface temperature uniformity tests should be completed during factory acceptance test (FAT), site acceptance test (SAT), and operational qualification (OQ). If significant changes to the heat transfer system are implemented, this test should be repeated as part of the change control.

Based on the studies presented, a typical shelf temperature mapping study for the 24 ft2 (2.2 m2) pilot unit used throughout these studies would consist of an empty chamber with five temperature probes per shelf. The indirect measurement method, using temperature probes or a suitable alternative, will provide greater accuracy and precision in the temperature measurement and reduce the need for repeat studies. The temperature probes should be placed at the geometric center of the shelf and within 1–3 inches (2.5–76 mm) of the four corners of the shelf.

The chamber should be evacuated to minimize the influence of the outside environment and more accurately demonstrate the shelf surface temperature uniformity and base line performance of the heat transfer system under no-load conditions. It should be noted that this evaluation provides a good baseline under no-load conditions. It is not the most challenging nor rigorous test in assessing the heat transfer system performance, including that of the shelf uniformity. The temperatures evaluated during the study should envelope the extreme low and high temperatures that will be run in the system, as well as one to three intermediate temperatures. The temperatures within the expected operating range should be selected to demonstrate the widest variability across the shelves. Part 3 will focus on approaches to the analysis of the data and justification for establishing an acceptance criteria from the studies presented here. 3

5