Introduction

Freeze-drying (lyophilization) is a drying technology widely used in the pharmaceutical industry [1]. It provides products with desired physicochemical properties, such as enhanced dissolution rates and bioavailability, and allows for the preservation and stabilization of thermolabile active pharmaceutical ingredients (APIs) and biologicals that have limited shelf life in solution [2,3,4]. It is important that product temperature during lyophilization remains below the characteristic glass transition/eutectic melting temperatures of amorphous/crystalline phases. Otherwise, above the glass transition temperature the amorphous phase turns oily and cannot support its own mass (collapse temperature), whereas above the eutectic melting temperature a liquid phase is formed, resulting in a massive loss of porous structure of the powdered material. Visually, the loss of structure of the dried matrix causes “cake collapse” in the freeze-dried product [4] and API recrystallizes into larger crystals of the same or other phases with drastic decrease of surface area.

Water is the most commonly used solvent in the freeze-drying method. However, scarce solubility of some hydrophobic and insoluble APIs in water severely limits the application of freeze-drying using aqueous solutions. This has led to the recent investigation of pure organic solvents or mixed organic–water solvents (further in the text termed «organic–water co-solvent systems») for use in freeze-drying processes. The use of a single-component organic solvent makes it impossible to prepare a stock solution containing water-soluble components, such as buffer salts or low-molecular mass carriers (mannitol, lactose, leucine, glycine, etc.). These are important solutions as they are used for producing solid dispersions to improve solubility for oral and/or intravenous drug delivery, and to enhance aerosolization properties of dry powder inhaler formulations [5]. In contrast, employment of organic–water co-solvent systems in the freeze-drying process facilitates manufacture of the stock solution by increasing the wettability and solubility of APIs, increasing sublimation rates, decreasing reconstitution times, and improving freeze-dried product stability [5,6,7,8].

To use organic–water co-solvent systems, it is necessary to find optimal compositions and concentrations of stock solutions, as well as optimal conditions for all stages of the freeze-drying cycle (freezing, annealing, primary drying, secondary drying). Moreover, most organic–water co-solvent solutions do not freeze in typical laboratory refrigerators because of their very low eutectic temperatures. Potential alternative methods for solidifying organic–water solutions are variants of cryogenic technologies, such as spray freeze-drying (SFD), spray-freezing into liquid (SFL) or thin film freezing (TFF) [9,10,11]. In these methods, the solution containing with API is sprayed either directly into a cryogenic liquid, usually liquid nitrogen (SFD, SFL), or over a cryogenically cooled solid surface (TFF). In the case of using some hydrophilic solvents (such as DMSO, ethanol, acetone, etc.) for mixed organic–aqueous solutions [12,13,14], the frozen species obtained by cryogenic technologies still have eutectic temperatures considerably lower than those available in typical pilot or laboratory freeze-dryers. Therefore, on attempted freeze-drying liquid phases containing significant amounts of associated water (solutions of water in the organic solvent) are formed. As a result, these formulations are nearly impossible to dry and unacceptable freeze-dried cakes are formed. The undesirable consequences include poor appearance, collapse, reconstitution of the samples, as well as a drastic degradation of the drug therapeutic activity.

To avoid formation of liquid phases, one needs to study the thermodynamics and kinetics of the processes related to all stages of freeze-drying. A phase diagram of the selected organic–water co-solvent system gives optimized parameters of a freeze-drying cycle only in the first approximation. Depending on the number of components (API, excipient), it is important to consider the relevant ternary/quaternary system in combination with corresponding binary systems (API-solvent, API-water, etc.). Further, one must also study the possibility of API/excipient hydrate (solvate) formation [15, 16]. In organic–water co-solvent systems, hydrates (water–TBA/ethanol/DMSO) [12, 13, 17] or clathrate hydrates (water–acetone/THF/1,4-dioxane) [14] can be formed on cooling. Clathrate hydrates are water-based crystalline compounds consisting of a lattice-like framework of hydrogen-bonded water molecules that form cavities, in which small (usually <0.9 nm) non-polar guest molecules of gases or volatile organics are incorporated [18]. Whereas clathrate hydrates of natural gases are widely applied for energy storage and transportation [19], there are limited reports on intentional use of clathrate hydrates for pharmaceutical applications [20,21,22,23].

Of common organic–water co-solvent systems, TBA–water is of particular interest and has been successfully employed in the commercial manufacture of CAVERJECT® Sterile Powder [24]. This co-solvent system is characterized by its low toxicity, its high eutectic melting point, thermal stability of hydrates that form on freezing, its low sublimation enthalpy and high vapor pressure. Further, the morphology of ice crystals formed on freezing accounts for a relatively low resistance of the partially dried solids to penetration of water vapor on sublimation. All these factors increase the sublimation rate of the TBA–water system [5,6,7,8, 24,25,26].

To develop new lyophilized API/protein formulations, one needs to study the phase diagram of the TBA–water–API ternary system. However, obtaining such a diagram—even in a limited concentration range—is complicated. The phase composition of the frozen TBA–water solutions does not correspond to that calculated from the binary TBA–water phase diagram (the formation of hydrates is not complete, a metastable hydrate is formed) and depends on the cooling rate [6, 17, 27]. Of further complication is the fact that crystallization of TBA and TBA hydrate is inhibited by the presence of excipients, such as sucrose and mannitol [24]. Even if, for simplicity, we neglect the presence of an API, a series of the intermediate phases can be formed in the TBA–water-excipient system, depending on the choice of excipient, the composition of the starting solution and the cooling conditions. As two extreme cases, one can propose two results, (a) ice Ih is the only crystalline phase, while the excipient, TBA and unfrozen water are amorphous, or (b) ice Ih, hydrates TBA·2H2O and TBA·7H2O (metastable) and a hydrate/solvate of the excipient are all crystalline, whereas both the excipient and TBA are amorphous; liquid water remains in this case. Potential consequences of these complex and poorly reproducible phase compositions are vial-to-vial variation in drying rates, collapse temperatures and residual TBA and water content in the final lyophiles [15, 24, 28, 29]. A large number of possible parameters must be varied empirically before optimum conditions can be found.

We supposed another organic solvent—water system, namely, the THF–water, to be not only easier to study, but also potentially more promising for practical applications. This is one of the best studied systems in relation to the chemistry of clathrate hydrates. The THF clathrate hydrate (cubic structure II (CS-II), Fd3m, a ~ 17.1 Å, molar ratio THF/water equal to 1:17) [14] crystallizes on cooling from a homogenous aqueous solution. The THF clathrate hydrate is stable up to fairly high temperatures (278 K, while the eutectic temperature in this system is 164 K) [14], thus making it possible to carry out freeze-drying across a certain concentration range at significantly higher temperatures.

Compared to the TBA–water–sucrose/lactose system [5,6,7, 17], when substituting TBA for THF, we have additionally substituted sucrose or lactose for glycine, a commonly used bulking agent in freeze-dried formulations. In contrast to sucrose or lactose, which tends to remain amorphous during lyophilization [30], glycine forms crystalline phases either directly on freezing, or as a result of subsequent annealing [30,31,32,33]. The main advantage of substituting the TBA–water–sucrose system for THF–water–glycine system is the possibility to prepare solutions with concentrations that give only 3, and in some cases only 2, crystalline (and no amorphous) phases on freezing or on freezing and subsequent annealing. It should also be noted that the physical state of the final lyophiles is the most important determinant of residual moister/organic solvent level [28, 30].

The aim of the present work was to study the phase diagram of the THF–water–glycine system along with the formation of the THF clathrate hydrate. The ternary phase diagram determination is essential in designing freeze-dried formulations and processes parameters for systems with a crystalline solute such as glycine [34]. In doing so, this study demonstrates parameter optimization for a freeze-drying cycle.

Experimental

Materials

Glycine (JSC “Reakhim”) was recrystallized to obtain pure α-form as described in [35]. Peroxide-free twice-distilled THF (JSC “VEKTON”) and distilled water were used as solvents.

Samples for low-temperature powder X-ray diffraction and DSC

Starting solutions glycine (5 mass%) in the THF/water co-solvent system (THF hydrate composition, 20 mass% of THF), THF–water solutions (20 mass%) (reference solution), aqueous glycine solutions (5 mass%) (reference solution). Freezing methods:

Method 1 Vials (Sci/Spec, B69308), 1 mL aliquots, air thermostat, 250 K (freezing time ca. 1 h).

Method 2 Vials (Sci/Spec, B69308), 1 mL aliquots, cryothermostat (KRYO-VT-05-02, TERMEX, Russia), 220 K (freezing time ca. 1 min).

Immediately after freezing the solutions (methods 1, 2) a few vials were broken under liquid nitrogen temperatures, in order to extract the samples, which were thereafter kept in liquid nitrogen until diffraction experiments (hereafter “frozen”: 1a, 2a). The other vials of these series were placed into an air thermostat at 250 K overnight for annealing (hereafter “annealed”: 1b, 2b).

Method 3 Small amounts of solution were splashed onto a copper plate cooled to liquid nitrogen temperature (freezing time ca. 1 s). The frozen solution was gently ground manually at liquid nitrogen temperature and was divided into two parts—one was kept in liquid nitrogen before diffraction experiments (hereafter “frozen”: 3a), a second was placed into an air thermostat at 250 K overnight (hereafter “annealed”: 3b).

Samples for drying rate determination and cryotemperature scanning electron microscopy (cryo-SEM)

Glycine solutions (5 mass%) in the THF/water co-solvent system were prepared by mass (5, 10, 15, 20 mass% of THF). An aqueous glycine solution served as the control. In each series, the solutions were put into ten vials (19 × 65 mm, 11 mL; Sci/Spec, B69308); the mass of the solution in each vial was 1.00 ± 0.05 g (vial fill depth of 5 mm, inner diameter standard and equal to 16.5 mm). Then the vials were tightly closed with a PTFE cap and placed into an air thermostat at 250 K overnight (12 h) before any further experiments.

Methods

Powder X-ray diffraction (PXRD)

PXRD experiments were aimed to: (a) identify the phases formed on cooling of glycine solutions in water and THF/water co-solvent system; (b) characterize final solid products. We used a Bruker D8 Advance diffractometer (λ = 1.5406 Å, tube voltage of 40 kV and tube current of 40 mA) equipped with an Anton Paar TTK 450 low-temperature chamber permitting work under vacuum down to 10−3 Torr. The vial with frozen solution was broken, the sample was gently ground in a mortar (all operations being performed at liquid nitrogen temperature) and placed onto a holder, which had been preliminary cooled to 130 K. Diffraction patterns were measured in the 150–280 K temperature range (2Θ scans from 5° to 45°, 0.02° step, approximately 8 min per pattern).

DSC

Samples of a frozen glycine solution were ground in a metal mortar cooled with liquid nitrogen, then loaded into a standard aluminum crucible (DSC-204 Netzsch, 25 μL) and also pre-cooled with liquid nitrogen. Sample mass varied from 0.16 to 2 mg. It is usual to measure the sample mass by weighing the crucible before and after loading the sample in it. Here, however, this was impossible because the sample must always be kept below 273 K. Therefore, we weighed the empty crucible, filled it with a sample, carried out the experiments, and only then weighed it again. The crucible filled with the sample was inserted into the DSC cell at 190 K. The heat flow sensor of the DSC showed an endothermic effect at that moment, proving that the crucible was below 190 K. The measurements were taken at different heating rates, 1 K min−1 (near 270 K), 2÷3 K min−1 (near 170 K) and 6 K min−1 (crystallization near 220 K), with a flow of pure Ar of 15 mL min−1.

Measurement of drying rate

Measurements of the drying rate were accomplished with a laboratory-scale freeze-dryer (one processing shelf, 25.0 × 35.0 cm, with temperature range of 240–350 K) (NIIC SB RAS, Russia) equipped with an organic solvent trap. A Convection-enhanced Pirani gauge [275 Mini-Convectron® (Granville-Phillips®)] was used to monitor the chamber pressure. The shelf was equilibrated at 268 K before vials were loaded and then placed under vacuum. Freeze-drying was carried out until the pressure dropped to P < 14 mTorr [DUO 60A (Pfeiffer Balzers)]. After that, the shelf temperature was increased to 303 K and held for 2 h. The pressure in the freeze-dryer was subsequently increased to P = 1 bar by filling it with dry nitrogen. Upon removal from the freeze-dryer, samples were stored in a vacuum desiccator over phosphorus pentoxide.

Blank test

Ten weighed vials with frozen (method 1) aqueous glycine (5 mass%) solution (1.00 ± 0.05 g per vial) were placed onto a preliminarily cooled shelf (268 K). Freeze-drying was carried out until the pressure dropped to P < 14 mTorr. The pressure in the freeze-dryer was then increased to P = 1 bar by filling it with dry nitrogen. The vials were removed from the freeze-dryer, closed, warmed to room temperature and weighed. Glycine: m calc. = 0.4843 g; m exp. = 0.4663 g.

Scanning electron microscopy (SEM)

The direct observation method by SEM was used to characterize the morphology of the frozen solutions. To extract a sample, the vial was carefully broken in the vessel with the liquid nitrogen temperature and the sample was placed on a brass holder, which had been already cooled to liquid nitrogen temperatures. Horizontal sample cuts were prepared that corresponded to a cross section of the cylindrical sample frozen in the vials. The holder was then placed into a staffing hole of an SEM TM-1000 (Hitachi) sample holder. It was determined experimentally that the sample temperature did not rise above 110 K during running.

Morphological examination of the lyophilized glycine samples was carried out with an EVO MA10 (Carl Zeiss) scanning electron microscope. The samples were mounted on a SEM stub with double-sided carbon tape and coated with gold to a thickness of about 8 nm with a JFC-1600 Auto Fine Coater (Jeol).

Results and discussion

Powder X-ray diffraction and DSC of the frozen solutions

The results of the low-temperature X-ray diffraction and DSC studies of the THF–water, glycine–water and glycine–THF–water solutions frozen under varying degrees of supercooling can be summarized as follows:

  1. 1.

    Cooling glycine solutions in THF/water co-solvent system (20 mass% THF) in vials (methods 1a, 2a) results in the crystallization of β-glycine (Fig. 1). Similar results have been observed for aqueous glycine solutions frozen at different cooling rates [31, 32]: DSC data indicated only the eutectic β-glycine–ice Ih. When the solution was quench-cooled by splashing it onto a copper plate that was cooled to liquid nitrogen temperatures (method 3a) (the highest cooling rate), glycine formed an amorphous freeze concentrate [31, 33] (Fig. 1).

    Fig. 1
    figure 1

    PXRD patterns (recorded at 150 K): “a”, “b” and “c”—powder diffraction patterns of frozen (a—method 1a; b—method 2a, c—method 3a) glycine solutions in THF/water co-solvent system (20 mass% of THF); “c” and “d”—powder diffraction patterns of frozen (c—method 1a; d—method 2a) THF–water solutions (20 mass% of THF), used as reference. Low intensity peaks (2ϴ range ~22.5°–26°) on patterns “b”–“e” correspond to ice Ih. The positions of the reflections of the THF hydrate (CS-II) and ice Ih are shown as ticks at the bottom. Asterisks—the strongest reflections of the β-glycine on the patterns “a” and “b”

    Full size image

Warming does not alter the powder diffraction patterns of samples frozen in vials (methods 1a, 2a), but some reflections of the low-temperature X-phase of glycine [31, 33] can be seen in the pattern of the sample prepared by method 3a and recorded at 210 K. On further warming (annealing), the reflections of β-glycine emerge in the PXRD patterns (Fig. 2). At the same time, the DSC curve of the sample prepared by method 3a shows a combination of thermal effects within the temperature range of 210–230 K (Fig. 3); similar results have been observed for the quench-frozen aqueous glycine solution [33]. Therefore, in solutions of the THF/water co-solvent system (method 3a), glycine undergoes a sequence of transformations: glassy/amorphous glycine → “X-phase” → β-glycine, as it does in the case of aqueous solutions [33].

Fig. 2
figure 2

PXRD patterns (recorded at 150 K) of the frozen (a—method 3a, b—method 3b) glycine solution in THF/water co-solvent system (20 mass% of THF). The positions of the reflections of the THF hydrate (CS-II) and ice Ih are shown as ticks at the bottom. Asterisks—the strongest reflections of the β-glycine

Full size image
Fig. 3
figure 3

DSC curves of the frozen glycine solutions (method 3; frozen—method 3a; annealed—method 3b) in THF/water co-solvent system

Full size image
  1. 2.

    The common feature of the DSC curves in all samples (methods 1–3) is a thermal effect at 269.1 K (melting of the eutectic ice Ih–β-glycine occurs at 269.9 K) [36] (Fig. 4).

    Fig. 4
    figure 4

    DSC curves of the frozen glycine solutions in THF/water co-solvent system (20 mass% of THF) (a, b—method 1; c, d—method 2; frozen—methods 1a, 2a; annealed—methods 1b, 2b)

    Full size image

A good agreement between the temperature dependences of unit cell parameters of ice Ih and THF hydrate obtained in our experiments with the literature data [37,38,39] (Fig. 5) proves the applicability of the experimental procedure used in this work.

Fig. 5
figure 5

Comparison of unit cell parameters of ice Ih (a) and THF hydrate (b) between 173 and 268 K obtained and in this work (triangles) and by Röttger et al. [37] (ice Ih), Ogienko et al. [38] (ice Ih) and Hester et al. [39] (THF hydrate)

Full size image

Phase diagram modeling

The results of the DSC experiments cannot be explained, assuming that the polythermal section “THF hydrate–β-glycine” of the state diagram of the THF–water–β-glycine ternary system is quasi-binary (i.e., assuming that β-glycine is an indifferent phase relative to the THF hydrate) (Fig. 6a). In the case of solution with composition of stoichiometry of the THF hydrate, the temperature of the congruent melting of the THF hydrate (278 K) is allowed as the maximum sample temperature during primary drying (Figs. 6a1, 7a1). If the composition of the solution corresponds to the THF hydrate–water subsystem (THF < ~20 mass%), the maximum allowable sample temperature during primary drying is the temperature of the eutectic melting of the THF hydrate–water system (271.7 K). The additional limitation is the eutectic temperature of the solute (β-glycine)–water subsystem (Figs. 6a2, 3, 7a2, 3).

Fig. 6
figure 6

I. Projections of the “ideal variant” (a) and the suggested model (b) of the temperature-composition state diagram for the “THF–water–β-glycine” ternary system (modeled in view of literature data [14, 36, 40]) on the composition triangle. Arrows in callouts designate the lowering of the temperatures of binary eutectics near the ternary peritectic and the ternary eutectic. Possible polymorphic transformations of glycine in the two- and three-phase fields containing liquid phases are neglected for clarity. Designations H—hydrate; Ih—ice Ih; Gly—β-glycine; l—solution; e 1, e 2, e 3, e 4—binary eutectics; E, E 1, E 2, P—ternary eutectics and ternary peritectic, respectively

Full size image
Fig. 7
figure 7

Polythermal sections of the “ideal variant” (a) and the suggested model (b) of the state diagram of the “THF–water–β-glycine” ternary system (a 1—“H–β-glycine” (line “1” on Fig. 6a); 2, 3—“H–water subsystem–β-glycine” (lines “2” and “3” on Fig. 6a); b 1—‘“THF–H subsystem”—β-glycine’ (line “1” on Fig. 6b); 2—“H–β-glycine” (line “2” on Fig. 6b); 3—‘“H–water subsystem”–β-glycine’ (line “3” on Fig. 6b)). Designations H—hydrate; Ih—ice Ih; Gly—β-glycine; l—solution; e 1, e 2, e 3, e 4—binary eutectics; E, E 1, E 2, P—ternary eutectics and ternary peritectic, respectively; T m(H), T m(Ih)—THF hydrate and ice Ih melting temperatures, respectively

Full size image

Therefore, we considered possible models of the state diagram of this system. Based on the experimental data and the suggested model (Fig. 6b), we conclude that the addition of the third component (β-glycine) changes the melting behavior of the THF hydrate. Melting of the THF hydrate in the THF–water binary system is congruent, whereas melting of the THF hydrate in the water–THF–glycine ternary system is incongruent. The thermal effect at 269.1 K corresponds to the four-phase peritectic reaction (T P): THF hydrate(solid) + β-glycine(solid) = ice Ih (solid) + solution. Thermal effects at 275–277 K (T H) are common to all samples and correspond to the melting of the THF hydrate (THF hydrate(solid) = solution) in the ternary system (in the THF–water system, the melting point of the THF hydrate is 278 K) [14]. Thermal effect at ~164 K observed in the DSC curves of the samples prepared by methods 1, 2 (Figs. 3, 4; Table 1) corresponds to the melting of the ternary eutectic (T E) in the THF–water–β-glycine ternary system (melting temperature of pure THF is 164.76 K) [40].

Table 1 DSC results for frozen glycine solutions in THF/water co-solvent system (20 mass% of THF)
Full size table

The effects observed at ~272 K (T 1) for samples prepared by methods 2, 3 correspond to the secondary melting [41] of the THF hydrate–ice Ih in the ternary system: THF hydrate(solid) + ice Ih (solid) = solution. The thermal effects registered within 255–265 K (T L) for the samples prepared by methods 2, 3 correspond to the transition of the imaging point (state point is defined as a position on the phase diagram representing a specific set of state variables) of the system from the ternary field THF hydrate–ice Ih–β-glycine to the binary field THF hydrate-liquor (shown by a dashed line in Fig. 7b1). The variation of the ratios of the thermal effects within 269–276 K for samples prepared by methods 2a/2b and 3a/3b indicates that on fast cooling the imaging point of the system diverts from the THF hydrate–β-glycine line. Besides, the presence of ice Ih reflections in the diffraction patterns of samples prepared by methods 3a/3b suggests partial evaporation of THF on freezing of the solution on the copper plate cooled to liquid nitrogen temperature. Therefore, each of the samples prepared by methods 2a and 3a can be considered as a combination of three states (the line THF hydrate–β-glycine; the subsystem THF hydrate–ice Ih–β-glycine; the subsystem THF hydrate–THF-β-glycine), so that the system is not in equilibrium (Fig. 7b1–b3).

The presence of only two thermal effects in the DSC curves (T P, T H) within the temperature range of 269–276 K, very weak (comparable to noise) reflections of ice Ih in the PXRD patterns of the samples prepared by method 1, as well as the ratio of thermal effects observed at 166 K and within 269–276 K indicate virtually complete (99.95%) formation of the THF hydrate under these conditions (Figs. 1, 3, 4; Table 1).

Three important conclusions can be made. (1) The THF hydrate forms almost quantitatively under conditions provided by typical laboratory freeze-dryers (method 1); (2) substitution of a single-component solvent (water) by a two-component one (THF–water) does not affect the polymorphism of glycine; (3) the limiting stage of freeze-drying is now the removal of the THF hydrate by sublimation at temperatures below the temperature of the four-phase peritectic reaction (269 K), but not the temperature of the congruent melting of the THF hydrate in the THF–water system (278 K).

Characteristics of frozen solutions

Freezing is a very critical step in the freeze-drying process that strongly influences the following primary and secondary drying stages and, ultimately, the quality of the final lyophilized product [42,43,44].

Cryo-SEM indicated obvious morphological differences of the frozen samples of glycine solutions, depending on THF content. Some typical images representing the most characteristic trends for each sample are selected and shown in Fig. 8. The examined surfaces were horizontal cuts made across the frozen cylinder formed in the vial. The presence of large (>100 μm) directional (dendrite type) ice crystals and the freeze-concentrated solute phase (frozen β-glycine–ice Ih eutectic) filling the space between the ice crystals was confirmed in the sample of aqueous glycine solution (Fig. 8a). These observations were in good agreement with previous results reported for nucleation of aqueous solutions of mannitol at low supercooling [45]. In the case of the sample with THF content corresponding to the THF hydrate composition (20 mass%) (Fig. 8c), individual crystals (<10 μm) are so small in size that they cannot be distinguished at the resolution used (×100–500). We consider this as an explanation of the fact that, according to [46], the onset of the hydrate formation in a solution with the stoichiometric water/THF mole ratio (17:1) occurs mostly inside the bulk liquid phase. For the sample with 15 mass% THF (Fig. 8b), the observed frozen structure corresponds to the close mixture of large THF hydrate crystals (truncated cubes/octahedrons) and of the crystals belonging to the solidus of the peritectic quadrangle (Fig. 7b3) THF hydrate–ice Ih–β-glycine (small and spherical).

Fig. 8
figure 8

Solvent crystal morphologies on horizontal cross section of frozen samples of aqueous glycine solution (a) and glycine solution in the THF/water co-solvent system (15 mass% (b); 20 mass% (c) of THF). Scale bar 500 μm

Full size image

The experiments on the pressure drop in the drying chamber as a function of THF content in the samples are illustrated in Fig. 9. These experiments have shown a significant shortening of the primary drying stage of frozen solutions of glycine in the THF–water co-solvent system as compared with aqueous solution: 135 min for THF–water (20 mass%) and 285 min for aqueous solution.

Fig. 9
figure 9

Pressure decrease in the drying chamber as a function of solvent composition of the glycine solution [line “1”—pure water (ice Ih is a single phase, removed by sublimation); line “2”—5 mass% of THF; line “3”—10 mass% of THF; line “4”—15 mass% of THF; line “5”—20 mass% of THF (THF hydrate is a single phase, removed by sublimation)]. The error bars are smaller than the symbols themselves

Full size image

It was demonstrated in [7] that the addition of small amounts of TBA (up to 10 mass%) to sucrose solutions not only significantly shortens the primary drying stage, but also increases the specific surface of the prepared sucrose samples, when compared with aqueous solutions. The shortening of the primary drying stage in the TBA–water co-solvent system is attributed to the formation of the TBA hydrate, the change (decreasing) in the freezing temperature affecting the morphology of the frozen solution and resulting in higher dispersiveness of the product, lowered sublimation enthalpy of the solvent and increased saturated vapor pressure [5,6,7,8]. Keeping in mind the estimated value of the sublimation enthalpy of the THF hydrate, derived from literature data [47,48,49], we attribute the shortening of the primary drying stage to a smaller required heat transfer to the samples for a given shelf temperature (∆sub H for THF hydrate and ice Ih: 2370 and 2830 kJ kg−1, respectively). It is well known for frozen aqueous solutions that the size of ice crystals must be large enough to obtain the shortest primary drying times (drying rates for slowly frozen samples typically greater than for fast frozen ones). Again, ice crystals must be smaller in size to produce a large specific surface area of the dried matrix in order to enable an easy desorption of unfrozen water from the pore surface to reduce secondary drying time [42, 44, 50]. In this case, however, it should be pointed out that the reduction of the sublimation enthalpy has a greater impact on reducing the drying time than the impact that a reduction in the size of crystallites has on increasing its duration. So, we suppose that in the case when freeze-drying process is rate limiting by the heat transfer (at very low product temperature), it would be of major importance for practical applications.

According to XRD data, in all samples obtained by freeze-drying, glycine was present as the metastable β-modification [51]. Visual inspection of the obtained sample revealed a stable freeze-dried cake lacking any signs of skin on the cake surface and collapsed inclusions near the bottom of the vial. At the same time, it should be noted that in some cases the upper part of the cake erupted from the vial (mass lost 7.1%) on freeze-drying glycine solutions in the THF–water co-solvent system (20 mass% of THF). In order to reduce sublimation rates, we carried out similar experiments at shelf temperature of 253 K. These experiments also have shown a significant shortening of the primary drying stage with a slight decrease in mass loss (4.9%). Furthermore, a significant change in the internal structure of freeze-dried cakes occurred on replacement of pure ice Ih with pure THF hydrate (20 mass% of THF) at the sublimation (removal) stage. According to SEM data (Fig. 10), it leads to a significant reduction of the particle size of glycine samples. We suppose that using the mannitol, trehalose or sucrose, which are able to form a mechanically strong cake [2, 30], can prevent the mass loss during primary drying. This will be the focus of future work.

Fig. 10
figure 10

SEM images of resulting cake structures of glycine samples, obtained by freeze-drying of glycine solution in the THF/water co-solvent system (20 mass% of THF) (a–c) and aqueous glycine solution (d–f). Scale bar a, d 50 μm; b, e 30 μm, and c, f 15 μm

Full size image

It should also be pointed out that THF is categorized as a Class 2 solvent in the ICH Harmonized Tripartite Guideline on Impurities: Guidelines for Residual Solvents Q3C (R5) [5, 52], whose use is limited in pharmaceutical products because of their inherent toxicity (TBA is categorized as a Class 3, includes no solvent known as a human health hazard at levels normally accepted in pharmaceuticals). It should also be pointed out that the pharmaceutical product NAB-paxlitaxel (nanoparticles albumin-bound paxlitaxel, one of the most important drugs in the modern treatment of metastatic breast cancer, Abraxane® [53,54,55]), is currently on the market that has been manufactured with the use of methylene chloride or chloroform, other Class 2 solvents [53,54,55,56,57].

The “injectable nanoparticle generator”, which spontaneously forms nanometer-sized doxorubicin-containing particles in aqueous solution, has been manufactured using dimethylformamide and methanol (Class 2 solvents) [58]. A cryogenic SFL process was developed to produce microparticulate powders consisting of an API molecularly embedded within a pharmaceutical excipient matrix; the samples have been manufactured using THF (Class 2 solvent) [59].

Conclusions

The THF hydrate crystallizes almost quantitatively on freezing from glycine solutions in the THF–water co-solvent system under conditions common for laboratory freeze-dryer. The substitution of a single-component solvent (water) with a two-component one (THF–water) does not affect the polymorphism of glycine. However, the substitution of sublimed ice Ih (pure water) for THF hydrate (co-solvent system of the composition corresponding to that of the clathrate hydrate) leads to a significant reduction of the particle size of glycine samples. Using a mixed solvent giving a clathrate on freezing is beneficial in terms of the significant shortening of the primary drying stage as compared with pure aqueous solutions and is of major importance for practical applications.