Temperature-modulated differential scanning calorimetry optimizes freeze-drying in drug formulation and preparation.

Although freeze-drying (lyophilization) is a standard process in the pharmaceutical industry for the manufacture of biologically active substances, it remains costly in terms of capital equipment and energy. New methods to optimize the key operating parameters (temperature, time, pressure, and concentration) could contribute significantly to product quality and process economics.

The freeze-drying process relies on the vapor pressure of ice. Even at –50 °C, ice sublimes, leaving a very porous, low-density cake containing the stabilized drug. The sublimation (drying) rate is temperature-dependent, and the maximum possible temperature during primary drying optimizes the process (1). The drying temperature selection also requires a knowledge of the physical characteristics of the formulation components, which are typically water, bulking agents, buffers or stabilizers, and the drug itself. The bulking agent plays an important role in stabilizing the physical structure of the frozen system and preventing its collapse. It can be amorphous or crystalline; the way that it interacts with the water and ice in the frozen solution defines the physical structure necessary for successful freeze-drying. The structure manifests itself in the form of transitions (e.g., glass transitions) that occur at specific temperatures and can be used to understand and optimize the process. This is important because the physical properties of the bulking agent (e.g., modulus, viscosity) can change by orders of magnitude at the glass transition. Analytical techniques that can accurately measure these structural changes and the temperatures at which they occur are critical in optimizing the freeze-drying process. One such technique is temperature-modulated differential scanning calorimetry or, more commonly, MDSC.

The Mechanics of MDSC
MDSC is an enhanced version of standard DSC, and is an established technique for accurate and reproducible measurement of crystalline or amorphous structure, for determining transition temperatures of materials (e.g., glass transition, crystallization), and for analysis of materials in which overlapping multiple transitions can occur at the same temperature (2). MDSC makes these measurements by subjecting the sample to a linear temperature ramp on which a small sinusoidal temperature modulation is superimposed. This allows MDSC to simultaneously measure the total heat flow associated with thermal transitions and the sample’s heat capacity. Signal subtraction supplies the kinetic, or time-dependent, component of the total signal. Thus, signals related to heat capacity (e.g., glass transitions) can be readily separated from kinetic signals (e.g., crystallization). Also, the dual heating rates permit enhanced sensitivity and resolution over those of standard DSC; these are important for detecting weak signals and improving resolution of overlapping transitions.

MDSC in Action
The ability of MDSC to analyze the complex transitions observed in frozen solutions is shown in Figure 1, where a 40% solution of sucrose in water was quench-cooled to –70 °C. A glass transition is seen in the heat capacity signal (–50 to –33 °C), while a crystallization peak appears in the kinetic component (–45 to –30 °C). The total signal, which is identical to that from standard DSC, shows only the sum of these two events and can be difficult to interpret. Crystallization of unfrozen water at the glass transition of these materials is well documented (3) and shows the maximum rate of ice formation occurring between –48 and –35 °C.

The physical state or structure of solutes after freezing determines not only the processing characteristics of the formulation, but also those of the final product, such as reconstitution, appearance, and stability. As shown above, the structure of a solute can be characterized by analysis of its glass transition (Tg). Accurate detection is important because within 5–10 °C of the Tg, solute physical properties can differ by several orders of magnitude. To further investigate this point, a series of MDSC experiments were conducted in which the solute was amorphous sucrose at 2.5%, 5%, 7.5%, and 10% (w/v), a typical range for freeze-drying formulations. Figure 2 shows the structure (heat capacity) of a 10% sucrose solution as it is cooled and then heated at 1 °C/min over the temperature range of the glass transition. Using the time-based derivative, step changes in the heat capacity signal appear as peaks, which simplify determination of the temperature midpoint of the transition. There are actually two step changes in heat capacity due to the glass transition of the sucrose, centered at –44 and –34 °C respectively. The reason for this is known, but exceeds the purpose of our discussion.

Figure 2 shows the structure immediately after the solution is frozen. Figure 3 shows the same structure, as measured during only the heating experiment, and compares it with the structure obtained after approximately 18 h at –40 °C, which is a typical freeze-drying temperature for this formulation. The structure changes, and the low-temperature step increases by 4 °C from –44.6 to –40.3 °C, the temperature at which the sample is held, while the high-temperature step near –34 °C is unaffected. Similar results were seen with 2.5%, 5%, and 7.5% sucrose concentrations. The low-temperature transition increase to the isothermal temperature of –40 °C is no coincidence.

During the 18-h experiment at –40 °C, it is believed that additional unfrozen water crystallizes and is therefore not available to plasticize some portion of the amorphous sucrose. The data from Figure 1 and the chart in reference 3 confirm that crystallization does occur at –40 °C. Since unfrozen water acts as a plasticizer, which tends to broaden the glass transition and lower its temperature, the conversion of unfrozen water to ice crystals is expected to cause a glass transition temperature increase.

Additional evidence supporting this theory is the rapid decrease with time in the heat capacity signal during the first 5–6 h at –40 °C, followed by a small but reproducible step after 7 h. This change is ascribed to the low-temperature transition, starting at –44 °C and advancing until it reaches the isothermal temperature, where the mobility of the molecules decreases. After this point, structure changes occur very slowly, and any further change in heat capacity (Cp) is due to decreasing mass as sublimation of the ice crystals continues.

The results show the ability of MDSC to detect structure changes at the glass transition and measure relative drying rates in the frozen solution. During the last 10 h of the isothermal segment at –40 °C and when compared with the 10% solute sample, the sample containing 2.5% solute dried 6.6 times faster. The drying time does not appear to be linear with concentration, as the samples with 5% and 7.5% solute dried 5.4 and 1.4 times faster than the 10% solute sample, respectively.

The Bottom Line
MDSC can separate complex transitions occurring in frozen solutions into their heat capacity and kinetic components. This offers solutions for the researcher or process engineer trying to optimize formulations or process parameters. These include the ability to measure

• the physical state or structure of the frozen solution accurately and precisely, permitting a more reliable selection of process temperatures;
• changes in structure with temperature and time, which should permit temperatures to be optimized as freeze-drying proceeds, thus reducing the time required and costs involved; and
• the effect of additives and concentrations on drying rates, which should reduce the total time required to develop reliable and cost-effective parameters for new or modified formulations.

Note: Modulated Differential Scanning Calorimetry (MDSC) is a registered trademark of TA Instruments, New Castle, DE.

References

1. Nail, S. I.; Schwegmani, J. J.; Kamp, V. Amer. Pharm. Rev. 2000, 3, 17–25.
2. Verdonck, E.; Schaap, K.; Thomas, L. C. Int. J. Pharm. 1999, 192, 3–20.
3. Roos, Y. H.; Karel, M.; Kokinii, J. L. Food Technol. 1996, 50, 95–105.