Abstract Digital twins (DTs) are emerging as the central enabler for an Industry 4.0 approach to pharmaceutical freeze-drying, promising real-time insight, closed-loop control and virtual validation of processes that are traditionally conservative, lengthy and material-intensive. This paper consolidates expertise from academia, equipment suppliers and biopharma manufacturers to define a five-level DT maturity model for lyophilization and to map the technologies, methodologies and regulatory concepts needed to progress through those levels. We first analyze the physical, mathematical and computational challenges unique to freezing, primary and secondary drying, and show how first-principles, CFD, coupled and probabilistic models can be combined with advanced PAT (wireless temperature probes, TDLAS, dP-dT mass-flow calculation, RGA, NIR) to create a continuously updated design space that is both equipment- and scale-aware. The ISO 23247 reference architecture is then translated into a lyophilization context to standardize data models and information exchange. Commercial platforms (Siemens xDT, Ansys Twin Builder, IMA Sentinel) are reviewed to illustrate implementation pathways, while emerging technologies-spray freeze-drying, RF/microwave heating and AI-enabled visual inspection-highlight the need for adaptable, hybrid mechanistic-ML twins. Regulatory guidance from FDA, ICH and ASME V&V 40 is synthesized into a risk-based verification and validation strategy that aligns DT credibility with product-quality impact. Finally, a road-map of actionable steps-spanning sensor selection, model calibration, uncertainty quantification, operator training, and data governance-is outlined for organizations to adopt during development, tech transfer, and commercial manufacture. Collectively, these recommendations provide a coherent framework for deploying fit-for-purpose, GMP-compliant DTs that shorten cycle times, enhance robustness and accelerate delivery of high-quality lyophilized products.

Diffusion properties of aqueous solutions of ubiquitous biopharmaceutical stabilizers (sucrose, sorbitol, glycerol) are studied by quasielastic neutron scattering (QENS). Molecular jump length and self-diffusion coefficients are determined for both water (Lw, Dw) and solutes (Ls, Ds) in 30 wt % solute/water mixtures at 298 K. Ds values scale inversely with the solute molecular size and follow Stokes-Einstein equation rank-order, while Dw values, unexpectedly, show an opposite trend, with the highest Dw and longest Lw in the sucrose solution followed by the sorbitol and glycerol solutions. The QENS data, combined with the literature NMR reports, indicate significant heterogeneity on both nm- and μm-length scales for the water/glycerol system, while water/sucrose appears to be homogeneous on the nm-length scale. The differences in the diffusion regimes in these solutions are rationalized via thermodynamics considerations.

Residual water in freeze-dried protein formulations plays a critical role in molecular relaxation behavior. In a previous study (Kumar et al., Mol Pharmaceutics 2025, 22, 4125-4136), we examined the impact of sucrose concentration on native structure retention in monoclonal antibody (mAb)-sucrose systems. The present work investigates how residual water content affects relaxation dynamics in this formulation system. Water affinity was first assessed via sorption isotherms across a range of water activities. There was an increase in residual water content as a function of mAb content in freeze-dried formulations. mAb-sucrose interactions were evident from detailed analyses of sorption data. There was a progressive decrease in the glass transition temperature with increasing water content. Dielectric spectroscopy was then employed to characterize the β- and γ-relaxation processes in the glassy state. While the β-relaxation time decreased with increasing water content, there was no significant effect on the corresponding activation energy. The effect of water content on the γ-relaxation was complex, with an increase in activation energies as a function of water content. In light of the influence of mobility on both physical and chemical stability, these results underscore the importance of controlling water content in optimizing the stability of freeze-dried mAb formulations.

This paper details the implementation of the ΔT and ΔP methodologies—recently introduced in the literature—for monitoring of mass flow during sublimation, demonstrating their application across multiple pharmaceutical companies. As a relatively new approach, this work is assessing its benefits, limitations, and potential enhancements. The mass flow of sublimated vapor per vial is the key parameter proposed to characterize the lyophilization process and for process transfer. Traditional methods like Tunable Diode Laser Absorption Spectroscopy (TDLAS) are expensive and difficult to implement in existing commercial equipment. The ΔT method measures the temperature difference between the inlet and outlet of the heat transfer fluid (usually silicone oil) circulating in the shelves. The sublimation process is endothermic, consuming heat and causing a slight temperature drop, which can be quantified to derive sublimation flow rates. The ΔP method utilizes the pressure difference between the chamber and the condenser, which occurs due to the vapor flow during sublimation. Both methods provide comparable data, validating their effectiveness in monitoring sublimation flow. The methods allow for monitoring of the primary drying segment of a freeze-drying process without the need for expensive equipment. They can be used to ensure that the freeze-drying cycle operates as intended, facilitating process transfers and scale-ups. The findings suggest that even older freeze-drying equipment can be retrofitted to utilize these methods, enhancing operational efficiency. The methodologies have been tested using process data by multiple pharmaceutical companies. The results indicate that the pressure drop and calorimetric methods are effective, cost-efficient alternatives to traditional monitoring techniques. They provide valuable insights into the freeze-drying process, helping optimize lyophilization cycle times. They also help establish equivalence between the primary drying endpoints at laboratory and industrial scales. These methodologies are suggested as a best practice for future technology transfers and process monitoring helping to ensure product quality, efficiency and consistency in biopharmaceutical manufacturing and contribute to the digital twin for lyophilization.

Water plays a critical role in chemical degradations, such as deamidation, in freeze-dried proteins. Two distinct patterns for deamidation in relation to water have been reported, that is a “hockey stick”-type behavior with a water-independent deamidation rate, followed by a sharp increase above a water content threshold, and an inverted bell-shaped profile. To understand the underlying mechanism, molecular dynamics simulations are employed to study the explicit water distributions around reactive sites for amorphous and crystalline insulin as well as amorphous IgG1. The simulated water distribution on the protein surface is first validated by successfully predicting water vapor sorption isotherms for both amorphous and crystalline insulin. The “hockey stick”-type behavior is explained by a water threshold level beyond which there are two (Asn-Gly sequence in IgG1) or three (Asn at the C-terminus in insulin) water molecules assisting the cyclization reactions. Regarding the inverted bell-shaped profile for amorphous IgG1, the initial decreases in deamidation rate with increasing water content at low water levels can be rationalized by a lower density and higher free volume of IgG1 at a lower water content. When the free volume exceeds a percolation threshold, the produced ammonia gas can easily diffuse away, lowering the back reaction rate and thus raising the overall reaction rate. The “free volume” mechanism can also be applied to the abnormal stability ranking orders of crystalline and amorphous insulin. The faster deamidation and dimerization rates in insulin crystals compared to amorphous insulin as reported by Pikal and Rigsbee are due to the lower density and higher free volume (above the percolation threshold) in crystalline insulin, assuming that dehydration of insulin crystals does not result in a major collapse of the crystal structure.

The study is part of investigations on relationships between water content, structure, and rates of chemical reactions in amorphous systems. This paper reports Asn deamidation of a hexapeptide in an amorphous liquid matrix, glycerol with water concentration of 1 to 30 wt%, at 50 °C. Using an amorphous liquid system allows focusing on the chemical and structural features of water effects, by minimizing the "molecular mobility" aspect. High-performance liquid chromatography (HPLC) is used to quantify both the loss of the parent compound and the accumulation of the cyclic succinimide intermediate and the hydrolysis products, Asp and iso-Asp. The rate constants for succinimide formation (k1) and succinimide hydrolysis (k2 and k3) are determined by fitting the HPLC data to specific kinetic models. The apparent pH of the solutions is confirmed to be independent of water content by using two orthogonal approaches. The experimental studies are complemented by molecular dynamics (MD) simulations of the hydrogen-bonding network around the Asn. This work reveals two water-content regions with distinct effects on deamidation. The first region shows a nearly constant k1 for water concentrations up to 8 wt%, whereas a significant increase in k1 with increased water content is observed in the second region above 12 wt% water. The water content threshold for the deamidation rate coincides with the spectroscopically determined thresholds for hydrogen bonding and water clustering in glycerol/water mixtures, as reported previously by a range of techniques including Raman spectroscopy. The study highlights relevance of hydrogen bonding and water clustering pattern for chemical processes including deamidation, and provides a basis for follow-up studies on the role of amorphous structure in deamidation in amorphous freeze-dried peptide and protein formulations.

High-resolution synchrotron X-ray powder diffraction has been utilized to detect mechanical stresses in frozen solutions, via analysis of the profiles of the diffraction peaks of ice. Increase in the width of the peaks (peaks broadening) reflects disruption of the crystal lattice, with contributions including a decrease in the crystallite size and an increase in microstrain. Frozen sucrose solutions (5 and 10 % w/v) were frozen at 100 K and then heated and annealed at -45 °C (228 K) and -15 °C (258 K), i.e. between the two apparent glass transition events in the freeze-concentrated solutions (Tg" and Tg'), and above the Tg', respectively. A decrease in microstrain and an increase in ice crystallite size were observed during annealing at -15 °C (above the Tg'), which is consistent with Ostwald ripening of ice crystals. Unexpectedly, and for the first time, an opposite trend was observed during annealing at the lower temperature of -45 °C, between Tg" and Tg'. To the best of our knowledge, this is the first report of an increased strain in the crystalline ice domains and of simultaneous size reduction during annealing of frozen aqueous solutions. Considering that the interaction with ice crystals may result in protein destabilization, the size and microstrain of ice crystallites can serve as markers of the freeze/thaw stresses on biopharmaceuticals. A practical implication is that a prolonged hold between Tg" and Tg' may increase stress imposed on protein molecules, i.e. a lower temperature is not always better for preserving biopolymers when freezing their aqueous solutions.

Sucrose is the most common stabilizer used in freeze-dried protein formulations. We have investigated, using several methods, the effect of monoclonal antibody (mAb) to sucrose weight ratio on the thermal, relaxation, and water sorption behavior of freeze-dried formulations. The influence of the sucrose content on the miscibility and retention of the native structure of mAb was also investigated. With decreasing mAb-to-sucrose weight ratio, the following effects were observed. Differential scanning calorimetry revealed a progressive decrease in the glass transition temperature of the formulation, while, based on dielectric spectroscopy, the α-relaxation time decreased, whereas both the β- and γ-relaxation times increased. The 1H T1 relaxation time of the antibody, determined by solid-state nuclear magnetic resonance spectroscopy, followed the same trend as the β-relaxation time. Finally, infrared spectroscopy indicated that the optimal retention of the native-like secondary structure of the antibody was achieved at a 4:1 mAb-to-sucrose weight ratio. At mAb-to-sucrose weight ratios of 1:1 and lower, there was no evidence of phase separation in the 20-50 nm scale. Taken together, the results provide new insights into the solid-state behavior of the antibody-sucrose system.

Two protein interaction peaks are observed in pharmaceutically-relevant protein (serum albumin) : disaccharide 1 : 1 and 1 : 3 (w/w) freeze-dried systems for the first time. In samples with a higher disaccharide content, the protein-protein distances are longer for both populations, while the fraction of the protein population with a shorter protein-protein distance is lower. Both factors would favor better stability against aggregation for disaccharide-rich protein formulations. This study provides direct experimental support for a "dilution" hypothesis as a potential stabilization mechanism for freeze-dried protein formulations. Two protein interaction peaks are observed in pharmaceutically-relevant protein (serum albumin) : disaccharide 1 : 1 and 1 : 3 (w/w) freeze-dried systems for the first time.

The purpose of this study was to investigate the mechanical stresses and strains acting on pharmaceutical glass tubing vials during freezing and thawing of model pharmaceutical formulations. Strain measurements were conducted inside of a laboratory-scale freeze-dryer using a custom wireless sensor. In both sucrose and trehalose formulations at concentrations between 5 % and 20 % w/v, the strain measurements initially increased before peaking in magnitude at temperatures close to the respective glass transition temperatures of the maximally freeze concentrated solutes, Tg'. We attribute this behavior to a shift in the mechanical properties of the frozen system from a purely elastic glass below Tg' to a viscoelastic rubber-like material above Tg'. That is, when the interstitial region becomes mechanically compliant at temperature above Tg'. The outputs were less predictable below 5 % w/v and tended to exhibit two separate peaks in strain output, one near the equilibrium melting temperature of pure ice and the other near Tg'. The peaks merged at concentrations between 4 and 5 % w/v where the largest strain magnitude was observed. The strain on primary packaging has traditionally been applied to evaluate the risk of damage or breakage due to, for example, crystallization of excipients. However, data collected during this study suggest there may be utility in formulation design or as a process analytical technology to minimize potentially destabilizing stresses and strains in the frozen formulation.

This paper reviews the structure and properties of amorphous active pharmaceutical ingredients (APIs), including small molecules and proteins, in the glassy state (below the glass transition temperature, Tg). Amorphous materials in the neat state and formulated with excipients as miscible amorphous mixtures are included, and the role of absorbed water in affecting glass structure and stability has also been considered. We defined the term “structure” to indicate the way the various molecules in a glass interact with each other and form distinctive molecular arrangements as regions or domains of varying number of molecules, molecular packing, and density. Evidence is presented to suggest that such systems generally exist as heterogeneous structures made up of high-density domains surrounded by a lower density arrangement of molecules, termed the microstructure. It has been shown that the method of preparation and the time frame for handling and storage can give rise to variable glass structures and varying physical properties. Throughout this paper, examples are given of theoretical, computer simulation, and experimental studies which focus on the nature of intermolecular interactions, the size of heterogeneous higher density domains, and the impact of such systems on the relative physical and chemical stability of pharmaceutical systems.

Changes in the protonation state of lyophilized proteins can impact structural integrity, chemical stability, and propensity to aggregate upon reconstitution. When a buffer is chosen, the freezing/drying process may result in dramatic changes in the protonation state of the protein due to ionization shift of the buffer. In order to determine whether protonation shifts are occurring, ionizable probes can be added to the formulation. Optical probes (dyes) have shown dramatic ionization changes in lyophilized products, but it is unclear whether the pH indicator is uniform throughout the matrix and whether the change in the pH indicator actually mirrors drug ionization changes. In solid-state NMR (SSNMR) spectroscopy, the chemical shift of the carbonyl carbon in carboxylic acids is very sensitive to the ionization state of the acid. Therefore, SSNMR can be used to measure ionization changes in a lyophilized matrix by employing a small quantity of an isotopically-labeled carboxylic acid species in the formulation. This paper compares the apparent pH of six trehalose-containing lyophilized buffer systems using SSNMR and UV-Vis diffuse reflectance spectroscopy (UVDRS). Both SSNMR and UVDRS results using two different ionization probes (butyric acid and bromocresol purple, respectively) showed little change in apparent acidity compared to the pre-lyophilized solution in a sodium citrate buffer, but a greater change was observed in potassium phosphate, sodium phosphate, and histidine buffers. While the trends between the two methods were similar, there were differences in the numerical values of equivalent pH (pHeq) observed between the two methods. The potential causes contributing to the differences are discussed.

Freezing and lyophilization have been utilized for decades to stabilize pharmaceutical and food products. Freezing a solution that contains dissolved salt and/or organic matter produces pure primary ice crystal grains separated by freeze-concentrated solutions (FCS). The microscopic size of the primary ice crystals depends on the cooling conditions and the concentration of the solutes. It is generally accepted that primary ice crystals size influences the rate of sublimation and also can impact physico-chemical behaviour of the species in the FCS. This article, however, presents a case where the secondary ice formed inside the FCS plays a critical role. We microscoped the structures of ice-cast FCS with an environmental scanning electron microscope and applied the aggregation-sensitive spectroscopic probe methylene blue to determine how the microstructure affects the mo-lecular arrangement. We show that slow cooling at-50 degrees C produces large salt crystals with a small specific surface, resulting in a high degree of molecular aggregation within the FCS. In contrast, fast liquid nitrogen cooling yields an ultrafine structure of salt crystals having a large specific surface area and, therefore, inducing smaller aggregation. The study highlights a critical role of secondary ice in solute aggregation and introduces methylene blue as a molecular probe to investigate freezing behaviour of aqueous systems with crystalline solute.

Objective The purpose of this paper is to re-visit the design of three steps in the freeze-drying process, namely freezing, primary drying, and secondary drying steps. Specifically, up-to-date recommendations for selecting freeze-drying conditions are provided based on the physical–chemical properties of formulations and engineering considerations. Methods and Results This paper discusses the fundamental factors to consider when selecting freezing, primary drying, and secondary drying conditions, and offers mathematical models for predicting the duration of each segment and product temperature during primary drying. Three simple heat/mass transfer primary drying (PD) models were tested, and their ability to predict product temperature and sublimation time showed good agreement. The PD models were validated based on the experimental data and utilized to tabulate the primary drying conditions for common pharmaceutical formulations, including amorphous and partially crystalline products. Examples of calculated drying cycles, including all steps, for typical amorphous and crystalline formulations are provided. Conclusions The authors revisited advice from a seminal paper by Tang and Pikal (Pharm Res. 21(2):191-200, 2004) on selecting freeze-drying process conditions and found that the majority of recommendations are still applicable today. There have been a number of advancements, including methods to promote ice nucleation and computer modeling for all steps of freeze-drying process. The authors created a database for primary drying and provided examples of complete freeze-drying cycles design. The paper may supplement the knowledge of scientists and formulators and serve as a user-friendly tool for quickly estimating the design space.

Stability of the majority of pharmaceuticals and biopharmaceuticals depends on acidity/basicity of the environment. In aqueous solutions, acidity/basicity is commonly expressed using proton activity scale, pH, while definition and experimental measurements of acid-base relationships in frozen and freeze-dried materials are less straightforward. The chapter starts with a brief summary of the current understanding of several critical aspects of pH and apparent acidity/basicity as related to freezing and freeze-drying, whereas the main part of the chapter is focused on areas which are underrepresented in the pharmaceutical literature, with both overview of the literature and previously unpublished data presented. In particular, the following topics are covered: (i) Hammett acidity function and pH-equivalent (pHeq); (ii) Factors which impact apparent solid-state acidity in lyophiles; (iii) Solid-state acidity and chemical instability of lyophiles; (iv) Freezing fundamentals: quasi-liquid layer, polarity of the freeze-concentrated solution, and the Workman-Reynolds potential. Potential directions for future studies in this field are also outlined.

Scale-up and transfer of lyophilization processes remain very challenging tasks considering the technical challenges and the high cost of the process itself. The challenges in scale-up and transfer were discussed in the first part of this paper and include vial breakage during freezing at commercial scale, cake resistance differences between scales, impact of differences in refrigeration capacities, and geometry on the performance of dryers. The second part of this work discusses successful and unsuccessful practices in scale-up and transfer based on the experience of the authors. Regulatory aspects of scale-up and transfer of lyophilization processes were also outlined including a topic on the equivalency of dryers. Based on an analysis of challenges and a summary of best practices, recommendations on scale-up and transfer of lyophilization processes are given including projections on future directions in this area of the freeze drying field. Recommendations on the choice of residual vacuum in the vials were also provided for a wide range of vial capacities. Graphical Abstract

Prediction of lyophilized product shelf-life using accelerated stability data requires understanding the temperature dependence of the degradation rate. Despite the abundance of published studies on stability of freeze-dried formulations and other amorphous materials, there are no definitive conclusions on the type of pattern one can expect for the temperature dependence of degradation. This lack of consensus represents a significant gap which may impact development and regulatory acceptance of freeze-dried pharmaceuticals and biopharmaceuticals. Review of the literature demonstrates that the temperature dependence of degradation rate constants in lyophiles can be represented by the Arrhenius equation in most cases. In some instances there is a break in the Arrhenius plot around the glass transition temperature or a related characteristic temperature. The majority of the activation energies (Ea), which are reported for various degradation pathways in lyophiles, falls in the range of 8 to 25 kcal/mol. The degradation Ea values for lyophiles are compared with the Ea for relaxation processes and diffusion in glasses, as wells as solution chemical reactions. Collectively, analysis of the literature demonstrates that the Arrhenius equation represents a reasonable empirical tool for analysis, presentation, and extrapolation of stability data for lyophiles, provided that specific conditions are met.

Protein structural changes during freezing and subsequent thawing are of great importance to a variety of biopharmaceutical applications. In this work, we studied the influence of non-ionic surfactants (polysorbate 20 and poloxamer 188) on protein structural changes during freeze and thaw using lysozyme as a model protein. Small-angle neutron scattering was employed to characterize protein structures in both liquid and frozen solution states. The results show minimal impact of polysorbate 20 on lysozyme structures during freeze and thaw using practically relevant concentrations. Polysorbate 20 used at 0.04% (w/w) completely prevents freeze-induced aggregation of lysozyme. Poloxamer 188 seems to interact with lysozyme; when applied at high concentrations (10% w/w), such interaction prevents protein crowding or close packing typically associated with freeze concentration. Despite such interactions, lysozyme aggregation is observed with 10% (w/w) of poloxamer 188 during freezing, although the aggregation is reversed upon thawing.

Water-to-ice transformation results in a 10% increase in volume, which can have a significant impact on biopharmaceuticals during freeze–thaw cycles due to the mechanical stresses imparted by the growing ice crystals. Whether these stresses would contribute to the destabilization of biopharmaceuticals depends on both the magnitude of the stress and sensitivity of a particular system to pressure and sheer stresses. To address the gap of the "magnitude" question, a phospholipid, 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC), is evaluated as a probe to detect and quantify the freeze-induced pressure. DPPC can form several phases under elevated pressure, and therefore, the detection of a high-pressure DPPC phase during freezing would be indicative of a freeze-induced pressure increase. In this study, the phase behavior of DPPC/water suspensions, which also contain the ice nucleation agent silver iodide, is monitored by synchrotron small/wide-angle X-ray scattering during the freeze–thaw transition. Cooling the suspensions leads to heterogeneous ice nucleation at approximately −7 °C, followed by a phase transition of DPPC between −11 and −40 °C. In this temperature range, the initial gel phase of DPPC, Lβ′, gradually converts to a second phase, tentatively identified as a high-pressure Gel III phase. The Lβ′-to-Gel III phase transition continues during an isothermal hold at −40 °C; a second (homogeneous) ice nucleation event of water confined in the interlamellar space is detected by differential scanning calorimetry (DSC) at the same temperature. The extent of the phase transition depends on the DPPC concentration, with a lower DPPC concentration (and therefore a higher ice fraction), resulting in a higher degree of Lβ′-to-Gel III conversion. By comparing the data from this study with the literature data on the pressure/temperature Lβ′/Gel III phase boundary and the lamellar lattice constant of the Lβ′ phase, the freeze-induced pressure is estimated to be approximately 0.2–2.6 kbar. The study introduces DPPC as a probe to detect a pressure increase during freezing, therefore addressing the gap between a theoretical possibility of protein destabilization by freeze-induced pressure and the current lack of methods to detect freeze-induced pressure. In addition, the observation of a freeze-induced phase transition in a phospholipid can improve the mechanistic understanding of factors that could disrupt the structure of lipid-based biopharmaceuticals, such as liposomes and mRNA vaccines, during freezing and thawing.