Computer-assisted sublimation

The sublimation step should be started when the matrix is in a completely frozen state. In order to ensure that the matrix will preserve its frozen state throughout the sublimation step, the computer should be provided with the collapse or eutectic temperature of the sample. Since the vast majority (99%) of samples will have a collapse rather than a eutectic temperature, the software will need to base the product temperature on some knowledge of, or means of computing that for the sample concerned. Such a frequency distribution of collapse temperatures arises from the fact that the glassy state, formed in the interstitial region of the matrix, is non-stoichiometric in composition, so the collapse temperature can vary from container to container depending on the particular frequency distribution. 

Based on the size (sample volume and number of vials) of the lot, the software must take into account that the calculated ideal product temperature may exceed the collapse temperature. 

The chamber pressure during the sublimation step (i.e. the primary drying process) has been found to be related to the product and shelf-surface temperatures [8] however, determining the shelf temperature required is more difficult as it depends on the nature of the heat transfer fluid used to control the shelf temperature and also on the particular design of the freeze-dryer. 

The sample container rests on a boundary layer at the top of the shelf surface. Such a layer is a region where the flow of heat transfer fluid is minimal or zero (i.e. the fluid is stationary). As a result, the sublimation step, which involves the transfer of heat from the fluid to the shelf surface, creates a temperature gradient across the boundary layer that depends on the thermal load exerted by the sublimation process and on the nature (viscosity, thermal conductivity and flow across the shelves) of the heat transfer fluid. The temperature gradient also depends on the number of shelves, their design and build, and on the pumping capacity of the circulation pump. These variables in turn depend on the size and particular manufacturer of the freeze-dryer, so the software used should include an input of data for the materials used, and for the dryer s design and build. 

The computer must determine the rate at which this step develops from the heat of sublimation of the ice in the sample and the uncrystallized water present in the interstitial region. Also, it should take into account the heat transfer coefficient of the sample container — in the case of a vial, not just a single container, but the frequency distribution of an entire lot of containers. From this statistical information, the computer can then determine the highest shelf-surface temperature where, based on knowledge of the batch size, the chances that the heat transfer coefficient of a container would result in a product temperature exceeding the collapse temperature and yielding a defective product are minimal. The frequency distribution of the heat transfer coefficients of the containers would also provide the computer with the information needed to extend the primary 

---