Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Received 13 September 2003; received in revised form 7 April 2004; accepted 8 April 2004
Abstract
This study was to investigate the effects of different freeze–drying factors on the rate of sublimation. The experiments were
carried out in a custom-built freeze–drying microbalance to accurately monitor the sample temperature and control the chamber
pressure. Twenty-four experiments were conducted based on a full factorial design by changing four factors: freezing rate (fast
freezing or slow freezing), chamber temperature (35, 0, or −35 ◦ C), chamber pressure (30 or 1000 mTorr), and the presence or
absence of an annealing process. Lactate dehydrogenase (LDH), a tetrameric protein, was selected as a model protein for this
study. The statistical analysis of the experimental results revealed that chamber temperature, analogous to the shelf temperature,
in this experiment system, had the greatest impact on the sublimation rate. High chamber temperature resulted in high sublimation
rate, regardless of the chamber pressure and thermal history of the sample. Chamber pressure was an important factor affecting
the sublimation rate. In addition, both chamber temperature and chamber pressure had significant impact on sample temperature
during freeze–drying. Annealing the samples was the most critical step to preserve good freeze–dried cake structure.
© 2004 Elsevier B.V. All rights reserved.
Keywords: Freeze–drying; Lyophilization; LDH; Sublimation; Microbalance; DSC
0378-5173/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.ijpharm.2004.04.011
96 J. Xiang et al. / International Journal of Pharmaceutics 279 (2004) 95–105
Freeze–drying is a time-consuming and expensive tors on the rate of sublimation by using a custom-built
process. Sublimation of ice during primary drying is freeze–drying microbalance. The four factors included
usually the longest stage in the whole freeze–drying freezing rate, shelf (chamber) temperature, chamber
cycle. In order to develop an economical freeze–drying pressure, and the inclusion of an annealing step during
cycle, it is crucial to minimize the process time, espe- freezing. Lactate dehydrogenase (LDH), a tetrameric
cially the time for primary drying. The ice sublima- protein, was selected as the model protein. LDH has
tion rate is the most important factor that determines been shown to be freezing and freeze–drying labile
the primary drying time (Pikal and Shah, 1990). Since since it tends to lose activity due to denaturation after
the major driving force for ice sublimation is the dif- freeze–thaw steps and freeze–drying (Nema and Avis,
ference in ice vapor pressure between the product and 1993; Izutsu et al., 1994, 1995; Jiang and Nail, 1998).
the condenser, the ice sublimation rate can be influ- The effects of the four freeze–drying factors on the ac-
enced by product temperature and chamber pressure tivity of LDH after freeze–drying were also examined
(Murgatroyd, 1997). For example, the sublimation rate in this study.
is increased as the product temperature is increased.
The product temperature is usually much lower than
the shelf temperature during primary drying due to 2. Freeze–drying microbalance
the energy consumption when ice is transformed to
vapor. The dried product that remains after ice subli- Pikal et al. (1983) designed a freeze–drying mi-
mation can form a barrier for the diffusion of water crobalance to study the sublimation behavior twenty
vapor underneath. This drying barrier, or resistance, is years ago. However, the freeze–drying microbalance
considered a major factor in reducing the sublimation they developed was not able to accurately measure the
rate of ice. In addition, the freezing process can influ- sample temperature or precisely control the chamber
ence the drying barrier structure. Freezing rate and an- pressure. In this study, a freeze–drying microbalance
nealing can affect ice crystallization, which will have (Austrian Research Centers, Seibersdorf, Austria)
impact on the size of the channels remaining in the was custom-built to provide better control and mea-
freeze–dried cake after ice is sublimated and thus the surement of the sample temperature and chamber
resistance to water vapor transport during the subli- pressure. A schematic diagram of this freeze–drying
mation. microbalance is shown in Fig. 1. The principal com-
The purpose of this study was to systemically inves- ponents of this freeze–drying microbalance included
tigate the effects of four important freeze–drying fac- a microbalance (BAL), a sample chamber, a con-
Fig. 1. Schematic plot of the freeze–drying microbalance. The principal components include (1) sample holder; (2) counter-balance; (3)
vacuum gauge; (4) vacuum valve; (5) vacuum pump; (6) sample chamber; and (7) condenser.
J. Xiang et al. / International Journal of Pharmaceutics 279 (2004) 95–105 97
Table 1
Freeze–drying conditions for the freeze–drying microbalance studies
Condition no. Sample chamber Annealing Freezing rate Pressure (mTorr)
temperature (◦ C)
1 −35 N Slow 1000
2 −35 Y Slow 1000
3 0 N Slow 30
4 0 Y Slow 1000
5 35 Y Slow 30
6 0 Y Slow 30
7 −35 N Slow 30
8 −35 N Fast 30
9 35 Y Slow 1000
10 0 N Slow 1000
11 35 Y Fast 1000
12 35 N Slow 1000
13 −35 Y Slow 30
14 −35 N Fast 1000
15 35 N Fast 30
16 0 Y Fast 30
17 −35 Y Fast 30
18 35 Y Fast 30
19 0 Y Fast 1000
20 35 N Slow 30
21 0 N Fast 1000
22 35 N Fast 1000
23 −35 Y Fast 1000
24 0 N Fast 30
J. Xiang et al. / International Journal of Pharmaceutics 279 (2004) 95–105 99
perature was maintained at a constant −80 ◦ C during then heated at 1 ◦ C/min to 80 ◦ C with or without an
the entire freeze–drying process. Annealing was con- annealing step. In an annealing step, the sample was
ducted by increasing the sample chamber temperature heated to −20 ◦ C at 1 ◦ C/min and held for 15 min,
at 1 ◦ C/min from −48 to −20 ◦ C, holding for 1 h, then cooled down to −48 ◦ C at 1 ◦ C/min and held for
and then cooling back to −48 ◦ C at 1 ◦ C/min, holding 10 min. The thermal profile was sampled every 0.5 s
for 0.5 h before evacuating the system. For each run, and the data were recorded by a personal computer
700 l LDH solution was transferred into a sample with TA data acquisition software and analyzed by
cup with an internal diameter of 10.89 mm. After each TA Universal Analysis software.
cycle, the freeze–dried product was reconstituted with
700 l of purified water and the activity was then
assayed. 4. Results and discussions
ature of this formulation without glycine crystalliza- trated in Fig. 5. The sample underwent an annealing
tion is around −50 ◦ C (Hey and Wang, 1999). There- process during freezing in which the sample chamber
fore, without annealing, the product temperature dur- temperature was increased from −48 to −20 ◦ C and
ing primary drying should be at least below −50 ◦ C then cooled to −48 ◦ C again. Ice sublimation was ini-
to avoid cake collapse. This would result in a long tiated by pulling the vacuum and increasing the sample
primary drying time and low efficiency. chamber temperature to 0 ◦ C. During the sublimation,
The DSC thermogram of LDH solution with an an- the sample temperature remained significantly lower
nealing step is shown in Fig. 4. The two endother- than the chamber temperature as a result of the heat
mic peaks at −1 and −6 ◦ C are the same as those absorption during sublimation. The dramatic increase
observed in Fig. 3. However, the small exothermic of sample temperature, due to the great reduction in
event observed in Fig. 3 did not appear after anneal- energy consumption, at the approximate time of 9.7 h
ing. Because most of the solute is crystalline glycine, signals the end of the sublimation process. The mi-
the highest allowable product temperature during pri- crobalance shows that the weight of the sample de-
mary drying after applying annealing in the freezing creased linearly during the sublimation.
phase would be the eutectic melting point of glycine at To study the factors affecting the sublimation rate,
−6 ◦ C, which would significantly reduce the primary twenty-four experiments were completed with various
drying time. combinations of freezing rate (fast or slow), annealing
(yes or no), chamber pressure (30 or 1000 mTorr), and
4.2. Freeze–drying microbalance study sample chamber temperature (35, 0 or −35 ◦ C). The
results of the experiments are summarized in Table 2.
The sample chamber temperatures, sample temper- There was no significant decrease in LDH activity af-
atures and weight losses in a typical freeze–drying ter freeze–drying under any of the conditions used,
cycle using this freeze–drying microbalance are illus- indicating the concentration of LDH (∼40 U/mL or
J. Xiang et al. / International Journal of Pharmaceutics 279 (2004) 95–105 101
0.3 mg/mL) used in the experiments was within the The experiment results were statistically analyzed
LDH stable concentration range. A similar result has by Statgraphics software based on the outcomes
been observed in a previous study (Hey and Wang, of four dependent variables: sublimation rate, aver-
1999). age sample temperature, cake structure, and enzyme
Table 2
The results of 24 freeze–drying experiments
Condition no. Sublimation rate (mg/h) Sample drying temperature (◦ C) Cake structurea Relative activity (%)
Fig. 8. Statistical analysis plot for cake structure (cake score: good cake = 10, loose cake = 5, and collapsed cake = 0).
of different freeze–drying conditions on the sublima- pressures. This result conflicts with other reported data
tion rate. In this study, the chamber temperature has (Searles et al., 2001) on the effect of annealing pro-
the lowest P-value in the analysis of variance for the cess on the sublimation rate. The different effects of
sublimation rate, revealing that the chamber temper- annealing on the sublimation rate could be attributed
ature has the greatest impact on the sublimation rate. to the different heat transfer patterns within different
The P-values of the chamber pressure and the anneal- freeze–drying equipment. For most freeze–dryers, the
ing process indicate the significant effects of these two vials were loaded on the shelves and most heat trans-
factors on the sublimation rate. However, the effect fer was carried out by conduction from the shelves to
of freezing rate on the sublimation is not significant. the vials. The direction of crystallization in the vials
As shown in Table 2 and Fig. 6, high chamber tem- was from the bottom to the top. For the freeze–drying
perature resulted in high sublimation rate, regardless microbalance, the sample cup was hung in the middle
of chamber pressure and thermal history of the sample. of sample chamber and heat transfer was carried out
With the same thermal history, increasing the chamber by convection and radiation. The direction of crystal-
pressure from 30 to 1000 mTorr also greatly increased lization in the sample cup was different from that in
the sublimation rate when the chamber temperature the vials, which could lead to different pore geometry
was 0 and 35 ◦ C, indicating that heat transfer from of the frozen bulk in the sample cup after annealing.
the chamber to the sample was a significant factor for Therefore, the annealing process in the freeze–drying
sublimation. However, when the chamber temperature microbalance could increase ice sublimation resis-
was −35 ◦ C, the trend of sublimation rate predicted tance by changing pore geometry of the frozen bulk
by the statistical modeling in Fig. 6 did not align well and shifted drying mechanism from MacKenzie types
with the experiment results shown in Table 2. With the I to II, which significantly reduced the sublimation
same thermal history, increasing the chamber pressure rate (MacKenzie and Luyet, 1963).
from 30 to 1000 mTorr at −35 ◦ C led to the decrease
of the sublimation rate. This was most likely attributed 4.2.2. Sample temperature
to the prohibiting effect of the high chamber pressure As revealed by the P-values shown in Table 3, the
(1000 mTorr) compared to the low ice vapor pressure chamber temperature again has the greatest contribu-
in the sample cup (about 168 mTorr at −35 ◦ C), re- tions to the change of sample temperature. The cham-
vealing that the chamber pressure played an important ber pressure and the annealing process can also sig-
role in the sublimation at low temperatures. nificantly affect the sample temperature. As shown in
Applying an annealing process during freeze–drying Table 2 and Fig. 7, higher chamber temperature re-
decreased the sublimation rate at all temperatures and sulted in higher sample temperature. Increasing the
104 J. Xiang et al. / International Journal of Pharmaceutics 279 (2004) 95–105
chamber pressure led to the increase of the sample which is critical in the early phase of drug formula-
temperature, due to the increased thermal convection tion and freeze–drying process development when the
at 1000 mTorr. Annealing the sample tended to pro- amount of available drug is limited. Since the heat
duce a higher sample temperature, which may result transfer pattern in the freeze–drying microbalance is
from the low sublimation rate in the sample. different from that in the conventional freeze–dryer,
the cake structure of the samples obtained from the
4.2.3. Cake structure freeze–drying microbalance could be different from
Based on the analysis of variance shown in Table 3, that obtained from the conventional freeze–dryer.
the annealing process is the most important factor in
determining the cake structure. The cube plot in Fig. 8
References
demonstrated that the annealing process resulted in
good cake formation as the result of crystallization of
Cao, Y., 2001. Endogenous angiogenesis inhibitors and their
glycine eutectic preserving the cake structure. Slow therapeutic implications. Int. J. Biochem. Cell Biol. 33, 357–
freezing and increasing the chamber temperature can 369.
improve the cake structure. Without the annealing pro- Dyer, D.F., Sunderland, J.E., 1968. Heat and mass transfer
cess, fast freezing resulted in cake collapse while slow mechanism in sublimation dehydration. J. Heat Transfer 90,
379.
freezing tended to form a loose cake. The retention of
Hey, J.M., Wang, D.Q., 1999. The effect of collapse on the stability
cake structure with slow freezing followed by primary of freeze–dried rFVIII and ␣-Amylase in KG-SF formulation.
drying at high chamber temperature was enhanced due Bayer Internal Report, unpublished data.
to the increased nucleation of glycine eutectic under Ivessa, E., 2000. From gene to protein therapy: how viruses can
slow freezing and the high sample temperature due to help. Trends Cell Biol. 10, 51.
Izutsu, K., Yoshioka, S., Kojima, S., 1995. Increased stabilizing
the high chamber temperature, encouraging the crys- effects of amphiphilic excipients on freeze–drying of lactate
tallization of glycine eutectic. dehydrogenase (LDH) by dispersion into sugar matrices.
Pharm. Res. 12, 838–843.
Izutsu, K.-i., Yoshioka, S., Terao, T., 1994. Stabilizing effect
5. Conclusion of amphiphilic excipients on the freeze–thawing and
freeze–drying of lactate dehydrogenase. Biotechnol. Bioeng.
43, 1102–1107.
The freeze–drying sublimation rate was studied by Jiang, S., Nail, S.L., 1998. Effect of process conditions on recovery
a serial of statistically designed experiments using a of protein activity after freezing and freeze–drying. Eur. J.
well-controlled freeze–drying system. This study con- Pharm. Biopharm. 45, 249–257.
Kuu, W.Y., McShane, J., Wong, J., 1995. Determination of mass
firms that chamber temperature has the greatest influ-
transfer coefficients during freeze drying using modeling and
ence on the sublimation rate. Increasing the chamber parameter estimation techniques. Int. J. Pharm. 124, 241–252.
temperature can greatly increase the sublimation rate. MacKenzie, A.P., Luyet, B.J. (Eds.), 1963. Effect of Recys-
Chamber pressure is also an important factor affect- tallization upon the Velocity of Freeze–Drying. Pergamon
ing the sublimation rate. Both chamber temperature Press, Munich.
Malafaya, P.B., Silva, G.A., Baran, E.T., Reis, R.L., 2002.
and chamber pressure has significant impact on the
Drug delivery therapies. II. Strategies for delivering bone
sample temperature during freeze–drying. The anneal- regenerating factors. Curr. Opin. Solid State Mater. Sci. 6,
ing process is the most critical factor in determining 297–312.
the retention of the cake structure in the formulation Murgatroyd, K., 1997. The freeze dyring process. In: Cameron,
used in this study. However, in the absence of anneal- P. (Ed.), Good Pharmaceutical Freeze–Drying Practice.
Interpharm Press, Buffalo Grove, pp. 1–58.
ing, the freezing rate can affect the final cake struc- Nema, S., Avis, K.E., 1993. Freeze–thaw studies of a model
ture. The freeze–drying microbalance is a convenient protein, lactate dehydrogenase, in the presence of cryoprotec-
and useful tool to study the sublimation rate during tants. J. Parenter. Sci. Technol. 47, 76–83.
a freeze–drying process. In addition, it can be used Pikal, M.J., Roy, M.L., Shah, S., 1984. Mass and heat transfer
to quantitatively evaluate the effects of freeze–drying in vial freeze–drying of pharmaceuticals: role of the vial. J.
Pharm. Sci. 73, 1224–1237.
variables in a freeze–drying process and thus to opti- Pikal, M.J., Shah, S., 1990. The collapse temperature in freeze
mize the freeze–drying cycle. The required amount of drying: dependence on measurement methodology and rate of
sample for the freeze–drying microbalance is small, water removal from the glassy phase. Int. J. Pharm. 165–186.
J. Xiang et al. / International Journal of Pharmaceutics 279 (2004) 95–105 105
Pikal, M.J., Shah, S., Senior, D., Lang, J.E., 1983. Physical Searles, J.A., Carpenter, J.F., Randolph, T.W., 2001. Annealing
chemistry of freeze–drying: measurement of sublimation rates to optimize the primary drying rate, reduce freezing-induced
for frozen aqueous solutions by a microbalance technique. J. drying rate heterogeneity, and determine Tg in pharmaceutical
Pharm. Sci. 72, 635–650. lyophilization. J. Pharm. Sci. 90, 872–887.