Abstract
The objective of this study was to examine the role of dimension and design of stoppers on the vapor transfer rate during the lyophilization process. Glass vials (100 cc, 28-mm neck diameter) containing mannitol solutions (4% w/w) were fitted with specially designed Teflon discs with circular holes (area ranging 0.12 to 41.80 mm2) instead of stopper vents, which served as the vent for water vapor transfer. The rate of vapor transfer across the vent, the temperature of the frozen product, and the vapor pressure inside the vial were monitored during the early primary drying phase. It was observed that the rate of vapor transfer from the vial increased as the cross-sectional area of the vent was increased. However, this increase tapered off as the area of the vent approached approximately 6 mm2 in the case when the shelf temperature was at 0 °C and 10 mm2 in the case when the shelf temperature was at 25 °C. Despite the higher accumulated pressure inside the vial and the corresponding higher temperatures of the frozen product, the rate of water vapor transfer across the vents was lower at the smaller vents, suggesting that the vapor transfer flow regime was under the choked conditions. Mathematical modeling using the observed water vapor pressure values inside the vial indicated that the flow pattern transitioned from the choked to non-choked pattern at a point where the ratio of pressure in the vial to pressure in the chamber, Pvial/Pchamber, was 2.5.
In another set of drying experiments using stoppers with various vent configurations (1-leg, 2-leg, and 3-leg) in glass vials (100 cc, 20-mm neck diameter) containing 4% w/w mannitol solutions, there was no statistical difference in the rate of drying among the three stoppers. The temperature of the frozen product/vial pressure profiles was also similar. This is due to the fact that the vent areas in all the cases were above ∼20 mm2 and coincided with the plateau region where mass transfer was not the rate-limiting factor across the vent areas. Partial blockage of the vent areas, even up to 50%, did not affect the rate of drying in any of the stoppers, whereas blockage of the vent areas of >75% did decrease the sublimation rate by 25%.
LAY ABSTRACT: The objective of this study was to investigate if the dimension and the design of stoppers may affect the moisture vapor transfer rate during the lyophilization process. The rate of vapor transfer across the vent, the temperature of the frozen product, and the vapor pressure inside the vial were monitored during the early primary drying phase in vials that were fitted with specially designed Teflon discs with circular holes (area ranging from 0.12 to 41.80 mm2) instead of stoppers. It was observed that although the rate of vapor transfer from the vial increased with the cross-sectional area of the vent, the increase tapered off as the area of the vent approached approximately 6 mm2 in the case when the shelf temperature was at 0 °C and 10 mm2 in the case when the shelf temperature was at 25 °C. Despite the higher accumulated pressure inside the vial and the corresponding higher temperatures of the frozen product, the rate of water vapor transfer across the vents was lower at the smaller vents, suggesting that the vapor transfer flow regime was under the choked conditions.
Studies with various stopper configurations (1-leg, 2-leg, and 3-leg) and vent areas also showed that as long as the vent areas are greater than the critical mass transfer restriction threshold then there should not be any difference in the overall drying behavior and that substitution of stoppers within the range should yield similar drying cycles.
Introduction
Freeze-drying, also known as lyophilization, is widely used for pharmaceuticals to improve long-term storage stability of heat- and hydrolysis-labile drugs as well as to provide easy handling during shipping and storage (1–3). The freeze-drying cycle essentially consists of three distinct phases: (1) Freezing of the solution, (2) primary drying or sublimation, and (3) secondary drying. Once the freezing of the solution is complete, the primary or sublimation drying begins, which is conducted under low chamber pressure conditions (e.g., 200 mTorr or lower) under which the sublimation of ice, as dictated by the ice/water-vapor equilibrium line of the phase diagram of water, takes place and the water vapor from the frozen matrix is transferred out of the vial, traveling into the headspace of the vial, through the vents of the closure, into the chamber, and eventually into the cold condenser, where it is condensed again as ice (Figure 1). Thus, frozen water from the vial is vaporized by sublimation and collected on the cold plates of the condenser. The sublimation is a phase change, requiring energy, which must be supplied as heat from the carefully controlled heated shelf. It is also a combined heat-mass transfer process in which both of the transfer phenomena must be balanced so that a sustained sublimation rate (mass transfer) prevails without collapsing or melting of the frozen mass due to accumulation of heat from the heated shelf (heat transfer). During the entire sublimation phase, the temperature of the frozen product should always be several degrees below the collapse temperature (Tc) in order to obtain a dry product with acceptable appearance. Factors influencing the rate of vaporization have been discussed extensively (4–7). The faster the heat can be supplied, the faster the drying proceeds, provided that the temperature of frozen product remains below its liquefying point, and sufficiently low pressure is maintained in the system by efficient vacuum pumps. If a sufficiently low pressure is not maintained, the temperature of the frozen product will rise, resulting in the partial softening or puffing of the product.
There have been studies (8–10) describing the impact of the key process parameters, such as shelf temperature, chamber pressure, heat ramps, and the duration of various cycle steps on the performance of lyophilization cycles. Even in robustly designed and well-controlled cycle runs, a portion or even an entire lot of vials sometimes exhibit unexpected and varying degrees of failure in drying. The assigned cause in such cases is often given as improper placement (inserted too far) of lyophilization stoppers in the vials, which may happen during the handling and transport of product-filled vials from a filling line to the lyophilizer chamber, because the stoppers at this stage are only placed snugly on the vial mouth and are relatively unstable. The accidental pushing of stoppers a little too much is possible, and in such situations obstruction in the water vapor transfer pathway may result in reduced rates of sublimation as well as increases in the temperature of the frozen product. In cycles with fixed shelf-temperature regimes, these vials may also lag behind the other vials, and upon progression of the cycle into stages of increased shelf temperatures, such as in secondary drying, these may result in partial or severe cake appearance defects.
The type of the vial/stopper assembly used has considerable influence on the outcome of the lyophilization process, especially at the manufacturing scale. In a review article (11), common problems encountered with lyophilization closures and desired properties of closures for lyophilized products are discussed. Cannon et al. (12) studied the impact of various vial design features such as glass type, vial diameter, vial bottom radius, and fill volume on the sublimation rates. The authors found that vial diameter offers the greatest impact on sublimation rates. Similarly, Pikal et al. reported that although the flow of water vapor experiences maximum resistance from the dried-product layer, the resistance due to the stopper vent and the chamber is also significant (13, 14).
However, the dependence of sublimation rate on the temperature of the frozen product and the water vapor pressure inside the vial has not been experimentally studied before. Such an attempt is made in this study. The rate of water vapor transfer (during the early part of the sublimation phase) was measured as a function of the dimension of the water vapor vent, various stopper configurations, and the placement/position of the stopper above the vial. In an innovative manner, the water vapor pressure inside the vial (above the cake) as a function of various vent areas was measured using a specially fabricated capacitance manometer assembly. These measurements were correlated with the temperature of the frozen product and the sublimation rate. A theoretical discussion on the mechanism of mass transfer is presented based on the experimentally determined average vial pressure and the vapor transfer rates (mass flux).
The specific objectives of this study were to
Correlate the vapor transfer rate, the average temperature of the frozen product, and the vial pressure during the early part of sublimation drying as a function of vent area
Compare the sublimation rates in various stoppers.
Materials and Methods
A. Correlation of Vapor Transfer Rate, Average Temperature of Frozen Product, and Vial Pressure during Early Part of Sublimation Drying as a Function of Vent Area
A.1. Preparation of Vials with Vents of Known Areas:
Specially fabricated Teflon discs that could be seated firmly on top of glass vials as shown in Figure 2 were used in this study as water vapor vents. Circular holes or orifices of areas ranging from 0.12 to 41.80 mm2 were precisely drilled in these discs. The discs were firmly placed on 100-cc glass vials having a 28-mm neck finish using Teflon tape to prevent any leaks. The holes served as the only water vapor transfer path from the vial to the exterior of the vial. In some discs, multiple holes of either the same size or different sizes were generated to give different effective vent areas.
A.2. Drying Rate Studies Preparation of Mannitol Solutions:
Twenty milliliters of 4% (w/w) mannitol solutions were 0.22 μm-filtered into the vials and the vials were accurately weighed and placed in a 12-ft2 shelf area in an external condenser lyophilizer (Vertis model Genesis 35L). The solution-filled study vials were surrounded by a cluster of other 100-cc empty vials to minimize the effect of the surrounding radiational heat input. Mannitol solutions instead of pure water were used in this study to avoid drying-related volume irregularities that generally happen in the case of sublimation of water alone. Mannitol solutions dry with well-maintained volumetric and symmetric structures, yielding reproducible drying kinetics.
Lyophilization Conditions:
The vials were frozen to −45 °C (with an annealing step at −15 °C for 3 h) for at least 6 h followed by initiation of the vacuum. The shelf temperature was raised to the target temperatures (0 or 25 °C) in 1.25 h and was maintained for an additional 3 h at this temperature. The chamber pressure was maintained at 200 ± 10 mTorr by controlling the nitrogen bleed valve.
Measurement of the Temperature of the Frozen Product:
The temperature of the frozen product was measured using thermocouple probes (Type K) placed inside the product near the bottom. Thermocouples were used in at least three vials for each set of holes. In some vials, thermocouples were placed such that the headspace temperature could be measured as shown in Figure 2.
Measurement of the Water Vapor Pressure Inside the Vials:
The vapor pressure inside the vial during drying was measured using a specially designed pressure probe assembly. It consisted of four capacitance pressure probes (MKS Instrument, model: Baratron) attached with stainless steel connectors (Figure 2) that were passed through the front plexiglas door of the lyophilizer into the chamber. The connectors were extended into the chamber to reach the vials placed on the shelves. Specially designed, 28-mm neck diameter, 100-cc glass vials were fitted with 2-in.-long stainless steel tubes in the vial wall such that the tube opened inside the vial just above the cake surface (Figure 2). The other end of the tube was connected to the stainless steel tube of the pressure probe assembly via a short (less than 2 in.) plastic tube. The offset of about 20 mTorr due to the tube length resistance was corrected from the measurement of vial pressure values to get the exact pressure in the vial from all the calculations (this offset of pressure drop due to the length of the tubing was obtained in experiments with and without the tubing). In each experiment, at least four vials were monitored for pressures using this assembly. The chamber pressure itself was also monitored/controlled by a separate capacitance manometer gauge.
A.3. Measurement of the Drying Rates:
To measure the rate of drying during the early part of the sublimation drying, the cycle was terminated at the end of 3 h; the vials were unloaded and weighed to calculate the amount of water lost per hour. In a separate set of experiments, the lyophilization cycles were terminated after the initial ramp of 1.25 h to the target temperature (0 or 25 °C). The amount of water lost during this period was measured and was subtracted from the total water lost from the studies where the drying was continued for 3 h at the target temperatures (0 or 25 °C). The drying rate per hour at the target shelf temperature was calculated by dividing the total amount in 3 h. Each value of the rate of drying represented an average from at least 10 vials and from multiple separate drying experiments.
The experiments were conducted for only 3 h in these studies to concentrate on the initial sublimation phase where drying occurs essentially from the top planer surface of the intact frozen cake. The maximum water lost in these experiments was less than 35% of the total volume (about 7 g out of 20 g) and the cake maintained its structure (no cracks or shrinkage to allow additional drying surfaces) to yield consistent and reproducible drying rate patterns in all the experiments. During this period, the temperature of the frozen product also remained relatively constant at its sublimation temperature. The drying rates were also obtained for unstoppered vials as well as vials seated with 28-mm diameter stoppers.
B. Comparison of the Sublimation Rates in Various Commercial Stoppers
To compare the rate of water vapor transfer across various stopper configurations, three different rubber stoppers of 20-mm diameter that are commonly used in commercial lyophilization were used as shown in Table II. The glass vials used in this study had a neck diameter of 20 mm instead of the 28 mm as described in the previous section due to unavailability of 28-mm stoppers in different configuration formats. The procedure to monitor the temperature of the frozen product and the vapor pressure inside the vials was the same as described earlier. The drying experiments were conducted for 3 h at a shelf temperature of 25 °C and a chamber pressure at 200 ± 10 mTorr. The vapor transfer rates were calculated as described earlier.
To simulate partial blockage of the stopper vent in the case of 2-leg stoppers, a set of stoppers was pushed down to the second notch, representing approximately 50% blockage. In some cases, the stoppers were pushed diagonally all the way down to represent more than 75% blockage of the overall vent area of the stopper (Figure 3).
Results and Discussion
A. Correlation of the Vapor Transfer Rate and the Temperature of Frozen Product/Vial Pressure during Early Part of Sublimation Drying as a Function of Vent Area
Table I shows the vapor transfer rate (i.e., average weight loss per hour during the first 3 h of sublimation drying), average frozen product temperature, and average vapor pressure inside the vial as a function of vent area under the condition of a shelf temperature of 25 °C and a chamber pressure of 200 mTorr. The vapor transfer rates were the average from multiple vials, which showed a variability of about 5% due to vial-to-vial differences in dimensions, thermal contact differences, and positioning of the vials in the chamber.
As shown in Figure 4, the vapor transfer rate increased with the area of the vapor vent from 0.12 to 41.80 mm2. However, the rate of increase became gradually smaller and tapered off at 10 mm2 and above. At a vent area of 10 mm2, an area equivalent to half that of the 28-mm diameter 2-leg stopper (vent area of 20.99 mm2), the rate was about 9% lower than the stopper itself. The rate of vapor transfer for the smallest vent (0.12 mm2) was approximately 87% lower than that of the stopper. The rate did not increase further even when the vent area was increased to 41.80 mm2, almost double the vent area of the stopper itself. The rates of vapor transfer for the effective area of approximately 20 mm2 created by multiple holes (10 × 1/16 in., 4 × 3/32 in. + 1 × 1/16 in., 2 × 9/64 in.) were similar and averaged 1.56 g/h. The rate of vapor transfer for the 28-mm 2-leg stopper was 1.68 g/h. It should be noted that the vents in the case of 28-mm diameter 2-leg stoppers are different in configuration than the vents created by circular holes in the Teflon disc.
A comparison of vapor transfer rates at 0 °C and 25 °C is shown in Figure 5. The vapor transfer rate profile tapered off at 6 mm2 in the case of 0 °C compared to 10 mm2 in the case of 25 °C, mainly because of the reduced rate of sublimation at 0 °C.
As the vent area became smaller, the vapor pressure inside the vial as well as the temperature of the frozen product increased (Figure 6). Despite the higher vial pressure gradient between the interior and the outside of the vial and the higher temperature of the frozen product, the drying rates were low for smaller vents. These observations are counter-intuitive to the expectation that the high vial pressure and high product temperature should result in faster drying rates. But the data clearly show that the drying rate became mass transfer-limited at smaller vents, and as a result of this, the vapor inside the vial continued to accumulate. The higher vapor pressure in the vial, in turn, caused a corresponding increase in the temperature of the frozen product (as governed by the vapor-pressure/temperature relationship). As the vent area was decreased to as small as 0.12 mm2, the vapor pressure inside the vial rose to almost 4104 mTorr (the phase diagram of water dictates that at around 4579 mTorr of vapor pressure, the temperature of the frozen ice approaches 0 °C). On the contrary, as the vent area approached 20 mm2 and above, there was relatively less accumulation of water vapor inside the vial and it was comparable to the exterior of the vial (chamber pressure).
In order to understand this observation of greater impedance to water vapor transfer at lower vent areas and plateauing as the vent areas were increased, the vapor transfer rates were evaluated in terms of diffusive and convective mass transfer mechanisms. To evaluate the flow mechanism, mass transfer flow equations were used to predict flow rates using the pressure differential observed at various vent areas. These equations are analogues to Bernoulli's principle and are applicable to flow across nozzles, venturies, or thin-plate orifices. The equations also assume a fully developed compressible flow in a horizontal plane without any frictional losses (15). In the vials/disc system under study, the water vapor flow starts from the wider planer surface of the cake, traverses through the narrower neck region, then through very thin orifice plates (Teflon discs) on top of the vial, and eventually opened up into semi-infinite space (chamber).
When the water vapor is transferred out of the vial through the vent or orifice on the Teflon disc, its flow through the opening may be choked or non-choked (15), resulting in the observation of the plateauing of the rates. Choked flow is a limiting condition that occurs when the mass flow rate does not increase with further decreases in the downstream pressure. For standard thin-plate orifice flow, the choked flow usually occurs when the ratio of the upstream pressure to the downstream pressure is 2 or above (15–18). The mass flow rate (Q, Kg/s) under such conditions is then described by the following equation: and the equation for the mass flow rate for non-choked flow is described as where C is the discharge coefficient; A is the vent area in m2; PVial and PChamber are the absolute upstream (vial) pressure and absolute downstream (chamber) pressure in Pascal, respectively; T is the absolute upstream gas temperature in Kelvin, k is the ratio of specific heat at constant pressure to specific heat at constant volume; ρ is the real gas density at PVial and T in kg/m3; M is the water vapor molecular weight in kg/kmol; R is the universal gas law constant in Pa · m3/(kmol · K); and Z is the gas compressibility factor at PVial and T.
In the above equations, most of the terms are standard physical constants; however, knowledge of exact values of discharge coefficient, C, PVial and PChamber, k, and compressibility factor Z is critical. In these calculations, the discharge coefficient is assumed to be equal to 0.6 based upon the available literature describing similar applications (19, 20). The chamber pressure is kept constant at 200 mTorr, however, the exact determination of pressure and temperature at the close vicinity of the orifice has practical limitations because the pressure probe could only be located near the center, at about 5 cm away from the vent, and the headspace temperature probe about 2.5 cm away from the orifice. In both the equations above, the mass flow rate is primarily dependent on the cross-sectional area A of the vent and the upstream pressure PVial, and only weakly (inverse square root) dependent on the temperature T.
Figure 7 shows the plot of water vapor transfer rates as a function of the ratio of vial pressure (PVial) to chamber pressure (PChamber). It exhibits a transition at point A, at a ratio of about 2.5, which approximates a boundary above which the conditions are in the choked flow regime. This is consistent with our earlier hypothesis that the mass transfer limitation was expected to diminish at higher vent areas. The PVial/PChamber ratio of 2.5 is similar to the ratio described by Searles (18) while studying the flow of vapor across the duct between the chamber and condenser of the lyophilizer. The PVial/PChamber ratio of 2.5 coincides with the vent area of approximately 10 mm2 as shown in Figure 8, suggesting a transition to the non-choked flow regime at higher vent areas. Based on this analysis, eq 1 was used to predict water flow rates for vent areas ≤10 mm2, while eq 2 was used in case of vent areas ≥10 mm2. It is evident that for lower vent areas, the resistance to mass transfer is significant and the flow regime resembles that of the choked flow condition.
Figure 8 shows the relationship between the calculated flow rates under choked and non-choked conditions and the observed flow rates as a function of vent area. The observed flow rates for choked conditions are lower than the predicted water vapor transfer rates due to overestimation of the Pvial. This is most likely because the pressure probe was located about 5 cm away from the orifice and the actual pressure at the orifice is expected to be much lower. Similar differences in the pressure values in the vicinity and away from the orifice have been reported by Eiamsa-ard et al. (21) through computational fluid dynamics simulations. The weaker correlation between the predicted and actual flow rates at lower vents may also be due to the unknowns associated with unique geometric features such as the shape of the entrance, the length of the orifice compared to its diameter (22), as well as the limitations in assigning proper values for C, Z, PVial, and TVial as mentioned earlier.
B. Comparison of Vapor Transfer Rates in Various Stoppers
Table II shows the comparison of water vapor transfer rates in commercially available stoppers with configurations of one, two, or three vents and with different effective total vent areas of stoppers. In spite of the differences in the configurations and total vent areas, there was no statistical difference (using analysis of variance single-factor analysis at a 95% confidence interval) in the vapor transfer rates among these stoppers even under a relatively aggressive shelf temperature condition of 25 °C. This was due to the fact that the vent areas of all these stoppers are in the plateau region of the vapor transfer drying rates as shown in Figure 4, and hence the vapor transfer rates were not limited by the mass transfer across the vents. In other words, all the water vapor sublimed from the frozen cakes, which was the same for all stoppers passed through the vents with similar impedance.
These stopper configurations offer varying degrees of operational flexibility. For example, in some high-speed filling machines, 3-leg stoppers are preferred over 2-leg stoppers because of the tendency of interlocking and machineability problems with 2-leg stoppers. In practice, it is always a question of whether a change of a stopper in a well set lyophilization cycle to a slightly different stopper (keeping the same number of legs) or a completely different stopper (having fewer or more legs) would affect the rate of sublimation drying and hence the outcome of the quality of the product. The results from these experiments suggest that there will be no difference in the outcome of the cycle by such interchange.
The data in Table II also show there was no decrease in the vapor transfer rate even when the 2-leg stoppers, which have the least total vent area, were pushed to the second notch to impart 50% blockage of the vent area. This again is consistent with the fact that the half-blocked vent did not fall in the mass transfer restricted curve of the drying profile. However, the vapor transfer rate was reduced by 25% when these stoppers were blocked further by more than 75% of their vent area. In practice, when a portion or even an entire lot of vials sometimes exhibit unexpected and varying degrees of collapse or meltback, the assigned cause in such cases is often given as improper placement (inserted too far) of the stoppers. Such blockage of the stoppers may happen during the handling and transport of product-filled vials from a filling line to the lyophilizer chamber. More than 75% blockage of the vapor mass transfer pathway in the example given here will result in a decreased rate of vapor transfer. These vials will also lag behind the other vials, and upon progression of the cycle into stages of increased shelf temperatures, such as in secondary drying, these vials will result in partial or severe cake appearance defects.
Conclusions
This study has shown that the water vapor transfer across the circular vents diminishes as the vent area is decreased and the drying rate profile follows choked and non-choked regimes. The transition from the choked to non-choked regime occurred at the vent area of 10 mm2 in the case when the lyophilization was conducted at 25 °C and at the vent area of 6 mm2 in the case when the lyophilization was conducted at 0 °C. Although the results of this study are specific to the vial/disc system and mannitol solutions used, the principles of vapor transfer during early sublimation drying should be applicable to other systems and solutions with suitable assumptions.
Studies with various stopper configurations and vent areas also showed that as long as the vent areas are greater than the critical mass transfer restriction threshold, there should not be any difference in the overall drying behavior, and substitution of stoppers within the range should yield similar drying cycles.
- © PDA, Inc. 2010