Lyophilisation (revision 32)

Lyophilisation, also called freeze-drying, is a dehydration process in which a solution is frozen and the ice is then removed by sublimation at a pressure below the triple point of water, leaving a porous solid whose volume approximates that of the original fill. It is the standard final step in the isolation of synthetic peptides after preparative purification, and the reason a research peptide is distributed as a dry cake in a sealed vial rather than as a solution.^[1]

The process is used because peptides in aqueous solution degrade by hydrolytic routes — deamidation of asparagine and glutamine, backbone cleavage, aggregation — whose rates depend strongly on molecular mobility and on the availability of water. Removing water and holding the residue in an amorphous glass suppresses those routes by orders of magnitude, converting a product with a shelf life of days at ambient temperature into one with a shelf life measured in years.^[2][3]

A lyophilisation cycle has three phases: freezing, primary drying in which ice is sublimed, and secondary drying in which water sorbed to the solid is desorbed. Each phase is bounded by a physical constraint. Freezing determines the ice-crystal structure and therefore the resistance the vapour must overcome later; primary drying must be conducted below the temperature at which the freeze-concentrated matrix loses rigidity; secondary drying must reach a residual moisture low enough for stability without over-drying a solid that may require a small amount of water for conformational integrity.^[4][5]

The finished cake is characterised by appearance, residual moisture, reconstitution time and the chemical purity of the reconstituted solution. None of these is visible from a purity figure alone, which is one reason a certificate of analysis that reports only area percent purity leaves the physical quality of the vial undocumented.^[6]

Water can pass directly from solid to vapour only at pressures below its triple point, which lies at 611.657 Pa and 0.01 °C. Above that pressure, warming ice produces liquid; below it, warming ice produces vapour. Every lyophilisation cycle therefore operates at a chamber pressure well beneath 611 Pa, typically between 5 and 20 Pa, so that heat supplied to the product drives sublimation rather than melting.^[1]

The driving force for sublimation is the difference between the vapour pressure of ice at the product temperature and the partial pressure of water vapour in the chamber. Because the vapour pressure of ice falls steeply with temperature, small changes in product temperature produce large changes in drying rate.

Values follow the standard formulations reviewed by Murphy and Koop.^[7] The table explains a practical asymmetry of the process: a cycle run with the product at −40 °C rather than −30 °C loses roughly two-thirds of its driving force, and a cycle run 5 °C warmer than intended may dry substantially faster while risking the structural failure described below.

Sublimation is strongly endothermic. The enthalpy of sublimation of ice is approximately 2.83 MJ·kg⁻¹, about 13% greater than the enthalpy of vaporisation of liquid water, so the shelf must supply a large heat flux for a modest mass flow. That heat must cross a gas gap between shelf and vial base, and the efficiency of that transfer — conventionally expressed as a vial heat transfer coefficient — is itself a function of chamber pressure, because conduction through rarefied gas falls as pressure falls.^[8]

Two consequences follow and recur throughout the article. First, product temperature is not a set point but a result: it emerges from the balance between heat supplied and heat consumed by sublimation, and it is lower than the shelf temperature throughout primary drying. Second, anything that changes the resistance of the dried layer to vapour flow changes the product temperature, which is why the freezing step — which sets that resistance — governs the behaviour of the two steps that follow.^[9]

Cooling an aqueous solution below 0 °C does not immediately produce ice. Nucleation requires supercooling, and in clean pharmaceutical solutions supercooling of 10–20 °C below the equilibrium freezing point is routine. The temperature at which nucleation happens to occur determines the number and size of the ice crystals formed: nucleation close to 0 °C produces few, large crystals and a coarse, low-resistance dried structure, whereas deep supercooling produces many small crystals and a fine structure that impedes vapour escape.^[9][10]

Because nucleation is stochastic, vials within one batch nucleate at different temperatures and therefore dry at different rates. This is a principal source of vial-to-vial variability in residual moisture and cake appearance, and it is the reason controlled-nucleation techniques — depressurisation, ice fog, or brief pressure cycling — have been developed to force all vials to nucleate within a narrow window.^[10]

As ice forms, solutes are excluded from the crystal lattice and concentrate in the remaining liquid. This freeze concentration continues until the residual phase either crystallises at a eutectic temperature or becomes so viscous that it vitrifies. The temperature at which the maximally freeze-concentrated amorphous phase undergoes its glass transition is written Tg′ and is a formulation property, not a process setting.

Figures are collated from the freeze-drying literature and vary with concentration and with the measurement method.^[1][4] The presence of sodium chloride is significant for peptide formulations because its low eutectic temperature drags the critical temperature of the whole system downward, forcing a colder and therefore longer primary drying step. A formulation carrying appreciable trifluoroacetate or phosphate salt from purification behaves similarly.

Annealing — holding the frozen product for a period above Tg′ but below the melting point — allows small ice crystals to grow at the expense of smaller ones and permits crystallising excipients such as mannitol to complete crystallisation before drying begins. Its benefit is a more uniform and less resistant dried structure; its cost is cycle time and, for some proteins, additional exposure to the ice-water interface.^[10]

Primary drying removes the ice, which for a typical aqueous fill is 90–98% of the water present. Its duration is set by the sublimation rate, which in the simplest treatment follows from an energy balance across the vial base: the mass flow equals the heat flow divided by the enthalpy of sublimation.

A worked estimate for a single 6R vial illustrates the magnitudes involved. Taking a vial cross-sectional area of 3.14 cm², a vial heat transfer coefficient of 20 W·m⁻²·K⁻¹, a shelf temperature of −10 °C and a product temperature of −30 °C, the heat flow is

Q = 20 W m^-2 K^-1 × 3.14 × 10^-4 m² × 20 K = 0.126 W

and the sublimation rate is

dm/dt = 0.126 W ÷ 2.83 × 10⁶ J kg^-1 = 4.4 × 10^-8 kg s^-1 ≈ 0.16 g h^-1

A 2.0 mL fill therefore requires approximately 12.5 hours of primary drying under these conditions, before any safety margin is added. Doubling the shelf-to-product temperature difference halves that figure, which is why the choice of critical temperature — and how close the cycle is permitted to approach it — dominates cycle economics.^[4][8]

Vials do not all behave alike. Those at the edge of a shelf receive additional radiant heat from chamber walls and door, sublime faster, and reach higher product temperatures; this edge-vial effect can amount to several degrees and is the usual reason a cycle validated on a laboratory dryer fails at production scale.^[8]

Two distinct structural failures bound primary drying. In a system whose freeze-concentrate crystallises, exceeding the eutectic melting temperature produces liquid, and the vial dries as a shrunken residue or a glassy shell — usually described on inspection as melt-back. In an amorphous system, exceeding a temperature a few degrees above Tg′ allows viscous flow of the freeze-concentrated matrix into the pores that the sublimed ice has vacated. The dried layer loses its structure, and the result is a shrunken, sometimes glassy or foamed cake. The temperature at which this becomes observable by freeze-drying microscopy is the collapse temperature, Tc, conventionally 1–3 °C above Tg′ for small-molecule and peptide systems.^[4]

Collapse is not necessarily a chemical failure. A collapsed cake may reconstitute and may assay within specification. Its documented consequences are elevated residual moisture, because the collapsed structure traps water; slower and sometimes incomplete reconstitution; and a cake appearance that fails a visual specification. For proteins, collapse has been associated with reduced storage stability, attributed to increased molecular mobility in the denser matrix.^[2]

Conservative practice therefore sets the target product temperature 2–5 °C below Tc and accepts the longer cycle. Aggressive cycles that run within a degree of Tc exist and are defensible when the critical temperature has been measured for the specific formulation rather than assumed from a table.^[4]

Primary drying ends when the last ice sublimes, and the transition is detectable because the vapour load on the condenser falls. Several signals are used. Comparative pressure measurement exploits the difference between a capacitance manometer, which reads total pressure, and a Pirani gauge, whose calibration depends on gas composition: as water vapour is replaced by nitrogen, the Pirani reading converges on the manometer reading, and the convergence marks the endpoint. Pressure-rise testing isolates the chamber briefly and measures the rate at which pressure recovers. Product thermocouples show a rise toward shelf temperature once no ice remains to consume heat, though a thermocouple perturbs nucleation in the vial it occupies and reads that vial rather than the batch.^[11]

The practical difficulty is that the last vials to finish are not the ones being monitored. Endpoint criteria are therefore normally applied with a hold period appended, and the batch is held under primary-drying conditions for a defined time after the signal is observed.^[11]

Once ice has gone, water remains sorbed to the solid — hydrogen-bonded to the peptide and to any amorphous excipient. Removing it requires desorption rather than sublimation, and desorption is driven by raising the product temperature well above the primary-drying value, typically to between 20 and 40 °C, while holding chamber pressure low.^[1]

Desorption kinetics are fast at first and then slow markedly, so the last fraction of a percent of moisture is disproportionately expensive in time. Secondary drying is consequently the step most often truncated, and truncation is a common explanation for a cake that meets appearance specifications but exceeds its moisture specification.^[2]

The heating ramp matters because the glass transition temperature of the partially dried solid rises as water leaves it. Early in secondary drying the solid is plasticised by residual water and its glass transition may be only slightly above the product temperature; heating too quickly can carry the product above the moving glass transition and cause the same collapse phenomenon seen in primary drying, at a much higher temperature. Conventional practice ramps shelf temperature at 0.1–0.5 °C per minute for this reason.^[4]

• Reconstitution of lyophilised peptides
• Karl Fischer titration
• Vial
• Vial filling and stoppering
• Peptide aggregation