Lyophilizer Shelf Temperature Mapping Part 1: Measuring
Batch uniformity and consistent critical quality attributes are vital to lyophilization as a process for product preservation. To achieve these goals adequate process control and reproducibility must be ensured.
Background on Shelf Temperature
Prior to processing product, the lyophilizer itself must be qualified to confirm the equipment is functioning properly and can be controlled within a specified allowable range. In addition to tests qualifying the independent lyophilizer performance variables, it is important to also quantify performance directly impacting batch uniformity. One such variable is the uniformity of the shelf surface temperature. Shelf temperature is a critical process parameter (CPP), and heat transfer occurs primarily at the shelf surface1.
The Food and Drug Administration has provided several forms of general guidance on lyophilizer qualification; however, there is little detailed guidance in the literature or regulatory documents regarding how to measure the shelf surface temperature2, 3, 4, 5. Making accurate measurements is an essential factor for achieving confidence in the results for reflecting the true shelf surface temperature.
A typical lyophilizer consists of a stainless steel pressure vessel referred to as a product chamber with temperature controlled shelves and a chilled condensing surface, either internal to the product chamber or external as a separate connected pressure vessel (see Figure 1). Components comprising the mechanical assemblies include a heat transfer fluid circuit that provides control of the shelf temperature and a vacuum system for reducing the pressure in the chamber and condenser. Functioning of these components to complete the process is orchestrated by an automated control system. A more detailed description of lyophilizer design and components has been well described over many years.6, 7, 8
The lyophilization process is commonly divided into three distinct parts. These are freezing, primary drying, and secondary drying. In freezing, the solvent, most commonly water, is first solidified. During primary drying the water is removed as subliming ice by reducing the chamber pressure and supplying heat. To remove any residual solvent that may lead to degradation during storage, residual solvent is desorbed by elevating the temperature. For all three parts the CPP of shelf temperature, chamber pressure, and time are controlled. The shelf temperature is important for each step of the process as the shelf is the primary route of heat flux, from the product during freezing and to the product during drying.
Chamber pressure is considered most critical during the primary drying segment when the sublimation of ice is being driven by the pressure differential between the product chamber and the vapor pressure of the solvent. The rate of sublimation is dictated by the quantity of heat supplied through the shelves. The time in each part is defined as the length of each soak at a constant target temperature or the time to ramp from one shelf temperature to another. Each part must be of a sufficient time to allow all the product containers to complete each step before proceeding to the next.
Commonly referred to as the shelf temperature, the fluid is monitored as the heat transfer fluid enters the manifold distributing the fluid to each of the shelves. The temperature is measured using a resistance temperature detector (RTD). This is inserted into a thermal well immersed in the heat transfer fluid path at the manifold inlet. The shelf surface temperature is dependent on the characteristics of the flow of heat transfer fluid through the shelves and the thickness of the stainless steel plates forming the top and bottom of the shelves.
To assure batch uniformity and process reproducibility, and therefore product with consistent critical quality attributes, each vial of the batch must experience the same thermal history. This is true regardless of the location on the shelf. Measuring shelf temperature uniformity through evaluation of the shelf surface temperature distribution is the first step in assessing the proper performance of the shelves as well as the heat transfer fluid circuit. Shelf temperature mapping is an initial assessment of flow of heat transfer fluid to all the shelves. Temperature mapping may detect a blockage or restriction in the manifold, heat transfer fluid hoses to each shelf, or within any section of a shelf. The test does not address if there is adequate capacity or flow through the heat transfer fluid circuit. This is because the test is conducted with no finished product containers or bulk trays containing water or a solution on the shelves and is therefore under no-load conditions.
One of the challenges in conducting such studies is to accurately measure the shelf surface temperature for comparing the shelf inlet temperature measured by the RTD. The conditions used for freezing and freeze drying create challenges in making good temperature measurements. This is partially due to influences of the environment within the product chamber. Studies were conducted to evaluate various methods for measuring shelf surface temperature. The goal of these studies was to determine which methods achieve a suitable level of accuracy and precision of the temperature measurement and best address the need for conducting a shelf temperature uniformity study.
Chamber Environment
The initial consideration when planning a shelf temperature mapping study is understanding the environment within the chamber9. Rambhatla conducted a study with the shelves fully loaded with material in vials to show uniform sublimation rates10. Though a valuable assessment of equipment capacity and overall performance, running under “full load” conditions introduces extraneous heat transfer influences and therefore is no longer simply evaluating the shelf surface temperature. Given the intention of evaluating the uniformity of fluid flow through all the shelves as a base line measurement prior to assessing the ability of the shelf system to act as a heat exchanger, it is preferred to run shelf temperature mapping studies with the shelves under no-load conditions. The performance of the shelves as a heat exchanger can best be evaluated by conducting specific studies to assess the performance of the entire freeze dryer as an integral system, through freezing, evacuation, and sublimation under “full load” conditions.
A second consideration in designing the shelf temperature mapping studies is the effect of chamber pressure. A study was conducted to compare the results when the mapping study was completed with the chamber at a reduced pressure (under vacuum), such as that used for the lyophilization process, and at or near ambient pressure. These studies were conducted using stand-alone wireless, direct measurement, and data storage temperature probes. These probes are a wireless device having a 1.4-inch (36-mm) diameter stainless steel base with an RTD secured through the center of the base. The temperature probes have a range from -85°C to 140°C with an accuracy of up to 0.1° and stability of 0.2°C and are designed specifically for measuring extremely low temperatures on surfaces. The temperature probes were placed at the corners and in the center of each shelf across four lyophilizer shelves. Measurements were made with the chamber at a reduced pressure of 200 microns (267 µbar) and at a slightly reduced pressure of 10–12 psia (0.7– 0.8 bar). These studies were conducted in a 24-square-foot (2.2 m2) internal condenser unit.
Figure 2 shows the average shelf surface temperature from the five temperature sensors (four corners and one center) across the four shelves when the system was at 200 microns (267 µbar) and at approximately one atmosphere. There was a clear top-to-bottom trend in the shelf surface temperature at both -50°C (see Figure 2A) and 50°C (see Figure 2B) when the system was at one atmosphere. This trend was minimized when the study was conducted with the system at a pressure of 200 microns (267 µbar).
The trend in the shelf surface temperatures when the study was conducted at one atmosphere demonstrates the chamber environment can influence the shelf surface temperature measurement. In this case, the influence may be due to the internal condensers which induce convectional flow of cold air passing by the shelves dropping to the chamber bottom. Given that the intention of this test is to evaluate the temperature range across the shelf surface of each shelf, eliminating potential influences from the environment is crucial; therefore, it is preferred to conduct shelf temperature mapping studies at reduced chamber pressures typically used for lyophilization.
Measuring Devices
One of the greatest challenges during shelf temperature mapping studies is the ability to accurately measure the actual shelf surface temperature. This requires good thermal contact between the shelf surface and the measuring device and preventing any influence of environmental conditions. Temperature measurement devices can be divided into two distinct groups: direct and indirect measurement. The direct measurement is a device which secures the temperature sensor, usually a thermocouple, directly to the shelf surface. The shelf temperature can be measured indirectly by attaching or embedding a temperature sensor in a device made of heat conductive material and resting the device on the shelf surface.
Various techniques can be used for the direct measurement method. These include a thermocouple taped to the shelf, a self-adhesive thermocouple, or a thermocouple held in place by a device such as a spring. If an adhesive is used, it must be able to hold the thermocouple tip in place throughout the temperature range to be evaluated. A spring assembly can work well in situations where the location being monitored is easy to reach when placing the device.
Multiple commercial devices are available for monitoring surface temperature using the indirect method. Some devices use a disc made from a heat conductive metal, such as copper, with an RTD embedded in the disc. These devices can also be readily assembled by taping a thermocouple to the top surface of a metal disc or placing it inside the disc through a hole drilled in the side to place the temperature sensor. There are also wireless devices commercially available which eliminate the challenges of multiple thermocouple wires running throughout the interior of the lyophilizer, such as the temperature probe described earlier.
The direct method has advantages as it provides a measurement of the actual surface temperature at a specific location. There are several drawbacks of the direct measurement method including the temperature sensors being difficult to place and difficult to keep in place across the operating temperature range imparted during the study. This typically requires adhesives and a heat conductive paste, which then needs to be removed from the shelf surface. However, for high capacity lyophilizers with shelve sizes of 20 square feet (1.85 m2) or larger, access to placing the temperature sensor is problematic and even a spring assembly can become difficult to place.
The main advantages of the indirect measurement are the ability to consistently place the devices in the same location and the elimination of any potential residues of adhesives or heat conductive paste. The main drawback of an indirect measurement device is the temperature measured reflects that of the device and not the shelf surface directly. The accuracy and precision of the measurement is influenced by the intimate thermal contact of the device to the shelf and can be influenced by the chamber environment. These influences can be addressed by a properly-established test method and a well-designed device.
Studies were undertaken to assess and compare common direct and indirect measurement methods. A small development size lyophilizer with an internal condenser having a capacity of 5 kilograms of ice was used for the study. The lyophilizer has a single refrigeration unit. The lyophilizer is fitted with a single 1-inch-thick (25-mm) stainless steel shelf, 1-foot (0.3 m) wide by 2 feet (0.6 m) deep, providing 2 square feet (0.18 m2) of shelf surface area. When operating at colder temperatures, the refrigeration is directed to chill the shelf preferentially and is then diverted to the condenser when the shelf is within 0.5°C of the target setpoint.
In the initial comparative studies, various types of fabricated devices were evaluated for their ability to measure the shelf surface temperature accurately and consistently. Making accurate and precise measurements are more challenging at sub-ambient temperatures, so measurements were made at -55°C and -15°C. An extreme temperature of -55°C was selected as it reflects the operating range of the lyophilizer and presents the greatest challenge in making good measurements by the various methods. An intermediate temperature of 15°C was selected because it is the midpoint of the range of 55°C and 25°C. All devices were compared to “taped” thermocouples. These “taped” sensors consisted of a 32-gauge type T thermocouple attached to the shelf using two 4-inch (102 mm) sections of duct tape arranged to form a cross with the thermocouple and heat conductive paste placed in the center under a 1-inch-by-1-inch (25-mm-by-25-mm) section of refrigeration insulation tape. All temperature sensors have a tolerance of +/- 0.5°C and were calibrated to within a Guard-Banding range of +/- 0.3°C, typical of proper procedures for calibrating temperature devices used for lyophilization11 (see Figure 3).
A spring assembly was constructed using a spring inside two overlapping pieces of plastic tubing with larger diameter end caps for physical stability. The spring assembly was longer than the distance between the lyophilizer shelves but short enough that when compressed it fit easily between the shelves. A thermocouple was placed on the shelf and the spring assembly was placed on top of the thermocouple and allowed to press against the shelf above to push the thermocouple tip down onto the top surface of the shelf below.
The screw disc device consisted of an approximately 2.5-inch (64-mm) diameter disc of stainless steel; a bolt with a nut at the end was placed on top of the disc. A thermocouple was secured between the bolt and the top of the disc. The bolt was long enough to reach the shelf above, and by unscrewing the bolt from the nut, downward pressure was exerted on the disc and thermocouple from the shelf above. A 1-inch-by-1-inch (25-mm-by-25-mm) section of insulation tape was placed over the thermocouple tip to prevent damage to the thermocouple, distribute the downward force of the bolt, and insulate the thermocouple from the device. This design was meant to improve heat transfer by forcing intimate contact between the disc and the shelf surface and the thermocouple tip and the disc. As an indirect method, this device eliminates the need for adhesive or heat conductive paste to be placed directly on the shelf. A second iteration of this device was tested with heat conductive paste between the top of the disc and the thermocouple tip.
A second indirect method consisted of a flat piece of metal with a thermocouple secured to the top with a 1-inch-by-1-inch (25-mm-by-25-mm) section of insulation tape. The metal size, shape, and composition were varied to see if surface area and shape affected the heat transfer between the shelf and the metal. These devices included a 2.25-inch-by-2.13-inch (57-mm-by-54-mm) plate of stainless steel, a 1.5-inch (38-mm) diameter disc of aluminum, and 2-inch, (51-mm) 2.5-inch (64-mm), and 3-inch (76-mm) diameter discs of stainless steel. The intention of these devices was to keep the device simple and easy to place while trying to improve the accuracy and precision with evaluating various types, sizes, and shapes of the device.
Various combinations of these devices were included in a series of studies to investigate their accuracy and consistency at measuring the shelf surface temperature at the target shelf inlet temperatures of -55°C and -15°C with the chamber pressure at 200 µmHg (267 µbar) or less. The shelf was controlled at the target temperature setpoint for at least 2 hours to allow the shelves and devices to reach a steady state temperature for an appropriate measurement. Data was analyzed for the last 60 minutes of the hold at each temperature and the average device temperature, standard deviation and the interquartile range (IQR) were calculated. The IQR was evaluated to assess the statistical dispersion, or spread of the data as the difference between the 75th and 25th percentiles, of the measured range. This data was compared to the average shelf inlet temperature over the same interval.
The results show all the measurement devices were more accurate at -15°C than at -55°C and none of the methods evaluated provided improved accuracy or precision as compared to the taped thermocouple method. Figure 4 summarizes the results at the target shelf temperatures of -55°C and -15°C. The spring assembly was the least accurate at both temperature conditions with an average difference from the shelf inlet of 7.6°C at -55°C and 2.3°C at -15°C. The screw disc performed comparably to the taped method at both temperatures with a 0.7°C difference between the two methods; however, at -55°C there was one outlier in the screw disk temperature suggesting the placement was not consistent in every study. The heat conductive paste with the screw disc increased the variability of the results and did not improve the accuracy or precision. At both low temperatures, there was no discernable difference between the different disc sizes or material of construction.
Based on the evaluation of the surface temperature measurement devices, the screw disc assembly could be a suitable device for measuring surface temperature from an accuracy and precision perspective when compared to the taped thermocouple method. The screw disc eliminates the need to clean up the adhesive and heat conductive paste from the shelf surface; however, the screw disc device was not able to address the concerns regarding ease of placement or keeping the thermocouples secured to eliminate study-to-study variability. None of the other constructed measurement devices were able to achieve an acceptable level of accuracy and precision.
Further studies were conducted to compare the taped thermocouple method to a commercially available surface temperature measuring device used in the previous study. For this purpose, a pilot size lyophilizer with an internal condenser and a capacity of 40 kilograms of ice was used. The lyophilizer has a single refrigeration unit. The lyophilizer is fitted with four shelves, each 2 feet (0.6 m) wide by 3 feet (0.9 m) deep providing 6 square feet (0.6 m2) of shelf surface area each, for a total of 24 square feet (2.4 m2). The locations of the temperature sensors were chosen so that the direct and indirect measurement devices would be positioned next to each other at the same location so they could be compared. All temperature sensors have a tolerance of +/- 0.5°C and were calibrated to within a Guard-Banding range of +/- 0.3°C.
The “corner” locations were 3 inches (76 mm) from the edges of the shelf. In the lyophilizer used, the shelves were constructed with internal “stay bars” comprised of solid stainless steel bars to direct the flow of heat transfer fluid in a serpentine pattern through the shelf. The stay bars also impart the structural strength for the pressure differential during lyophilization and distribute the forces when fully seating the stoppers in the vials at the completion of the process. These stainless steel bars are welded on center at 6-inch (152-mm) intervals; therefore, 3 inches (76 mm) from the edge would be near the center of the fluid flow path inside the shelf. For the “center” location, the sensors were aligned along the center of the shelf from side-to-side and front-to-back which would be directly over the center stay bar.
Four consecutive runs were conducted with the temperature sensors removed and replaced after each study. The same locations were used in each study; however, individual sensors were not placed in identical locations for each study to eliminate any calibration bias. Fresh tape was used in each study to secure the thermocouples. Target shelf inlet temperatures of 50°C and 50°C with the chamber pressure at 200 µmHg (267 µbar) or less were assessed. The target shelf temperature was held as a temperature soak for at least three hours to allow the devices to reach a stable measurement: these temperature probes needed the extra hour to stabilize due to the mass of the probe. Data was analyzed for the last 60 minutes of the soak at each temperature and the average device temperature reading was calculated. This average reading was compared across all 20 measurement devices for the same time period.
The taped thermocouple method used in these studies was the same as the tape method in the previous studies described earlier. The indirect method used in these studies consisted of the wireless temperature probes. The wireless temperature probe is comprised of a 1.4-inch (36-mm) diameter stainless steel base and an RTD secured through the center of the base. The temperature probe is supplemented with a thermal pad that adheres to the base and improves the thermal conductivity between the probe base and shelf surface. Due to the improved heat transfer, the thermal pad diminishes the time required to reach stable conditions and ensures more accurate and precise measurements, particularly under reduced pressure conditions.
Figure 5 is a comparison of the range for the 20 temperature sensors during each of the four consecutive studies. Over the course of the four studies, the direct method thermocouples were highly variable across the 20 locations. This was typically due to 1 or 2 of the measurements being different from the rest and the location of the 1 or 2 different measurements varied in each study. This type of variability indicates the thermocouple was not in intimate contact with the shelf surface and was not measuring the true shelf surface temperature. The temperature probes consistently reflected the true shelf surface temperature as compared to those measured by the taped thermocouples. Data of study 3 presented in Figure 5 demonstrates the accuracy of the thermocouples can be equivalent or slightly better than the temperature probes if all 20 thermocouples read consistently throughout the study.
At the end of each study, the thermocouples were inspected for evidence the tip was not in intimate thermal contact with the shelf. If such an observation were made, the data from that thermocouple could be eliminated from the analysis as the reading would not represent the true shelf surface temperature. In all four studies, no thermocouples were excluded based on observing the tip not in contact with the shelf surface. This suggests it is not always possible to observe the thermocouple tip rising off the shelf surface when under the tape.
The taped thermocouple method has the potential to provide a slightly better measurement of the actual surface temperature measurement as compared to temperature probes when all the thermocouples remain in place throughout the entire study. Yet, the improvement is less than the error inherent in the measurement method. When compared with the ability of the temperature probes to provide consistent measurements across multiple studies, there is no meaningful advantage to using the taped thermocouple method. The inconsistency of the direct measurement is of particular concern due to the conflict with the objective of the shelf temperature mapping study.
Conclusion
The intention of a shelf temperature mapping study is to evaluate uniformity of the shelf surface temperature. This will ensure uniform heat transfer fluid flow through the entire shelf bundle. The direct measurement method is more variable from run to run when the thermocouple tip does not remain in intimate thermal contact with the shelf surface. In order to account for this variability, multiple runs would be needed to show the different measurement was an outlier and not true variability in the shelf surface temperature.
When choosing a measurement method, primary consideration should be given to the accuracy and precision of the measurement and the ability to detect the true shelf surface temperature. The nominal difference between the direct thermocouple and indirect temperature probe method accuracy does not justify one method over the other. Therefore, elements such as consistency of the measurement by the temperature probe sensing device and the efforts in properly and consistently positioning the temperature sensors for completing the test should be considered. The greatest benefit of a wireless indirect measurement such as the temperature probe is the confidence of a consistent measurement which eliminates the need for repeat studies and investigations into failed results.
The studies conducted demonstrate the care needed when designing a shelf temperature mapping study. The intention of the shelf temperature mapping study is to assure acceptable uniformity of the shelf surface temperature and thereby confirm uniform heat transfer fluid flow through the shelf manifold. Therefore, a clean, dry, and empty chamber provides the least complexity for a shelf temperature mapping study by eliminating any potential sources of variability other than the heat transfer fluid circuit and shelf performance. Further, conducting the test at a reduced pressure, in a range typically used for processing provides a clearer picture of the performance of the shelves without influence from the chamber environment.
The measurement device used in conducting a shelf mapping study should be selected based upon its ability to measure the shelf surface temperature with sufficient accuracy and precision. The studies conducted demonstrate that a thermocouple taped directly to the shelf surface can be the most accurate method. Yet, this method has challenges in placement, achieving adequate precision, and difficult cleanup. The screw device also showed acceptable accuracy and precision but only eliminated the need for cleanup because it still has similar challenges with precision and placement as the taped thermocouples. The screw device would also be prohibitively difficult to place on a shelf with a depth of more than 3 feet (0.9 m). A commercially available surface temperature measurement device, such as the temperature probes, showed accuracy comparable to the taped thermocouple with excellent precision from study to study. The temperature probes were also simple to place and required no cleanup. Based on these results, the commercially available wireless surface temperature measurement devices provide the best all-around performance for conducting shelf temperature mapping studies.
