Dynamical in-situ observation of the lyophilization and vacuum-drying processes of a model biopharmaceutical system by an environmental scanning electron microscope

Abstract

The paper discusses the real-time monitoring of the changing sample morphology during the entire lyophilization (freeze-drying) and vacuum-drying processes of model biopharmaceutical solutions by using an environmental scanning electron microscope (ESEM); the device’s micromanipulators were used to study the interior of the samples in-situ without exposing the samples to atmospheric water vapor. The individual collapse temperatures (T_c) of the formulations, pure bovine serum albumin (BSA) and BSA/sucrose mixtures, ranged from −5 to −29 °C. We evaluated the impact of the freezing method (spontaneous freezing, controlled ice nucleation, and spray freezing) on the morphologies of the lyophiles at the constant drying temperature of −20 °C. The formulations with T_c above −20 °C resulted in the lyophiles’ morphologies significantly dependent on the freezing method. We interpret the observations as an interplay of the freezing rates and directionalities, both of which markedly influence the morphologies of the frozen formulations, and, subsequently, the drying process and the mechanical stability of the freeze-dried cake. The formulation with T_c below −20 °C yielded a collapsed cake with features independent of the freezing method. The vacuum-drying produced a material with a smooth and pore-free surface, where deep cracks developed at the end of the process.

Introduction

Many pharmaceuticals and biopharmaceuticals have proved to be unstable in aqueous solutions. Such materials are usually stabilized by drying, with lyophilization (freeze-drying) being the most common method; other approaches to be considered include, for example, vacuum-drying (Mattern et al., 1999, Mattern et al., 1997) and spray-freeze-drying, an unconventional freeze-drying technique that involves the generation of small droplets, flash freezing, and drying (Ishwarya et al., 2015, Wanning et al., 2015). As regards the numerous freeze-dried pharmaceuticals, we can consider, for example, blockbuster protein drugs, including Enbrel and Remicade; small molecular weight antibacterial drugs, exemplified by cephalosporins (such as Convenia); and many viral vaccines (e.g., those employed to treat measles, mumps, and rubella). Vacuum-drying facilitates the production of BOTOX, which has found use in multiple aesthetic and medical applications. A typical lyophilization process comprises 3 steps: freezing, primary drying (ice sublimation), and secondary drying (removal of unfrozen water). Freezing embodies a critical stage in the lyophilization procedure, as both the efficiency of the drying and the quality of the final product are directly influenced by the freezing conditions. In particular, the processes that exploit controlled ice nucleation, where freezing is initiated at a pre-selected temperature (usually slightly below the equilibrium freezing point), could be beneficial due to the shorter production times enabled by the faster ice sublimation (Geidobler and Winter, 2013) or better recovery of the product potency (Fang et al., 2018).

Besides the empirical approaches to freezing and lyophilization, there are numerous attempts, including some based on our research, to designate and separate the freezing-related stresses that might affect the stability and reactivity of frozen and freeze-dried compounds: low temperature (Bhatnagar et al., 2007, Kasper and Friess, 2011), freeze-concentration (Bhatnagar et al., 2007, Heger et al., 2005, Kania et al., 2014, Kasper and Friess, 2011, Krausko et al., 2015, Ondrušková et al., 2018; Takenaka and Bandow, 2007), ice formation (Bhatnagar et al., 2007, Kasper and Friess, 2011), freezing-induced acidity change (Gomez et al., 2001; Govindarajan et al., 2006; Heger et al., 2006, Imrichová et al., 2019, Krausková et al., 2016; Sundaramurthi et al., 2010; Vetráková et al., 2017), and polarity variations (Heger and Klán, 2007). Recently, mathematical simulations were applied to predict the optimal freezing protocol (Arsiccio et al., 2020) and the morphology of freeze-dried products (Capozzi and Pisano, 2018). In order to identify optimal freezing conditions, the morphologies of freeze-dried cakes prepared under diverse freezing processes are commonly evaluated. Methods for evaluating the morphological features of freeze-dried materials include, e.g., determining the specific surface area (Rambhatla et al., 2004), X-ray tomography (Foerst et al., 2019, Pisano et al., 2017), mercury porosimetry (Hottot et al., 2007), optical coherence tomography-based freeze-drying microscopy (Mujat et al., 2012), and scanning electron microscopy (SEM). Of these, the last-named technique embodies probably the most common tool to evaluate the morphologies of freeze-dried cakes: The SEM has been employed, for example, to examine the impact exerted on the cake morphology by controlled ice nucleation via different methods (pressurization-depressurization (Awotwe-Otoo et al., 2013, Iyer et al., 2016, Kawasaki et al., 2018), vacuum-triggered ice nucleation (Oddone et al., 2017), and ice nucleation through partial vacuum followed by its quick release (Izutsu et al., 2014)), to analyze the pores in freeze-dried cakes produced via various freeze-drying cycles (Capozzi et al., 2019, Liu, 2006, Overcashier et al., 1999), to visualize the crystallization of active pharmaceutical ingredients and excipients (Murase et al., 1991, Ogienko et al., 2014, Shalaev et al., 1996), to evaluate the morphologies of vacuum-dried pharmaceuticals (Mattern et al., 1999, Mattern et al., 1997), and to predict the resistance of lyophilized products to vapor flow (Arsiccio et al., 2019, Arsiccio et al., 2017, Grassini et al., 2016, Pisano and Capozzi, 2017).

In such SEM studies, a common underlying assumption is that the morphology (e.g., porosity) of the freeze-dried cake accurately characterizes that of the ice crystals in the frozen sample, if no collapse occurs during the primary and secondary drying. While this assumption is intuitively logical, there remains a lack of research reports on direct comparison of the morphology of a frozen material with that of the freeze-dried cake. Understanding the morphologies of ice crystals is essential for both practical applications, such as developing optimal freezing conditions for the manufacturing of freeze-dried products, and understanding the impact of freezing on the stability of protein molecules.

A conventional SEM requires low pressures (<10⁻² Pa) to avoid undesirable scattering of the electron beam due to collisions with the atoms and molecules of gases (Reimer, 1998). The low pressure requirement confines this technique to the observation of samples that are stable in vacuum conditions; the majority of water-containing materials cannot be studied with the method. Ice and frozen aqueous solutions are commonly examined via SEMs equipped with a cryo-stage at temperatures below −60 °C (Baker et al., 2007, Baker et al., 2003, Baker and Cullen, 2003; Barnes et al., 2002a, Barnes et al., 2002b, Barnes et al., 2003, Barnes and Wolff, 2004, Blackford et al., 2007, Chen and Baker, 2010, Cross, 1969, Cullen and Baker, 2001, Dominé et al., 2003, Erbe et al., 2003, McCarthy et al., 2013, Obbard et al., 2003, Rango et al., 2000, Rango et al., 1996, Rosenthal et al., 2007, Wergin et al., 2006b, Wergin et al., 2006a, Wergin et al., 1998), as imaging at higher temperatures leads to difficulty maintaining the SEM vacuum due to excessive ice sublimation (Cullen and Baker, 2001). In contrast, temperatures typical of the freezing and primary drying steps of the lyophilization procedure are −50 to −10 °C. Another problem relates to observing dielectric samples when a higher energy electron beam is used: Such samples become negatively charged, and their observation is hindered. This effect is usually eliminated in a sample surface coated with a thin conductive layer (Goldstein et al., 2003). The layer nevertheless interferes with the actual monitoring; in this context, for example, it modifies the chemical composition or masks fine morphological details (Fourie, 1982, Jones, 2012) and might prevent investigation of the processes occurring in the microscopic chamber during the visualization.

To enable direct observation of samples in their native states, the first commercial environmental scanning electron microscope (ESEM) was introduced around 1980 (Danilatos, 1981). Differential pumping enables an ESEM to maintain a high difference between the very low pressure near the electron gun and the relatively high pressure in the specimen chamber. The primary electron beam is formed and focused in a high-vacuum environment and then transferred through pressure-limiting apertures to the specimen chamber, where it travels only a very short distance to reduce the primary-beam scattering in the gas (Danilatos, 1988). Still, a portion of the primary electrons are scattered by the gas, leading to a decrease in the detected signal-to-noise ratio and a lower resolution of the ESEM. Conversely, the gas molecules are ionized due to collisions with the signal electrons emitted from the sample; such procedures then result in efficient signal amplification (Neděla et al., 2018, Neděla et al., 2015) and also compensation of the negative charge on the sample surface. Thus, electrically non-conductive samples do not require conductive coating (Stokes, 2008). Due to the higher pressure (up to thousands of Pa) of the water vapor in the specimen chamber, biological (Vlašínová et al., 2017), polymeric (Bertóková et al., 2015), or other water-containing samples (Navrátilová et al., 2017) can be examined without being dehydrated (Stokes, 2003). In wet samples, controlling the sample temperature and air/vapor pressure in the chamber of the ESEM allows the thermodynamic equilibrium to be established or perturbed to facilitate investigation of the phase transitions (e.g., condensation, evaporation, sublimation, and crystallization) (Donald, 2003, Krausko et al., 2014, Neděla et al., 2020, Schenkmayerová et al., 2014, Tihlaříková et al., 2013, Yang et al., 2017). The ESEM was used for studying the ice nucleation processes (Varanasi et al., 2010, Zimmermann et al., 2008, Zimmermann et al., 2007) and growth of ice crystals (Pedersen et al., 2011), observing metastable liquid water droplets below the pressure of the triple point (Chen et al., 2017), visualizing the brine on the ice surface (Imrichová et al., 2019, Krausko et al., 2014, Vetráková et al., 2019), and examining the sublimation of the frost flowers (Yang et al., 2017) or pure ice (Nair et al., 2018) in the microscope’s chamber. It was shown previously that an ESEM can be used to observe the whole lyophilization process under pharmaceutically relevant conditions (Meredith et al., 1996); this is the only study discussing lyophilization or vacuum-drying performed via an ESEM we are aware of.

This paper presents dynamical in-situ observation of the drying step in the lyophilization of a model protein BSA, both pure and mixed with sucrose, inside the chamber of an ESEM. The solutions were subjected to freezing under various conditions (spontaneous freezing, controlled ice nucleation, and spray freezing), whereas the drying conditions were kept constant. While the first two freezing methods are commonly used in conventional freeze-drying, the third one embodies the emerging technology of spray-freeze-drying (Ishwarya et al., 2015). The effect of the freezing method on the progress of the lyophilization and the structure of the resulting lyophile was thoroughly examined (in this article, the terms structure and morphology are used interchangeably). Unexpectedly, and contrary to the underlying assumptions proposed in previous SEM studies of freeze-dried materials, the lyophile morphology can be significantly altered during the ice sublimation, even at temperatures well below the collapse temperature (T_c). Therefore, studying a final freeze-dried material might not be enough to yield relevant data on the morphology of the ice crystals formed during the freezing step. This observation could have a far-reaching impact on both the design of the freezing conditions for various freeze-dried products and the fundamental understanding of the protein/ice interaction and destabilization of the protein molecules by ice. Finally, the vacuum-drying process was also monitored, revealing remarkable differences in the sample morphology.