Spray Drying ASDs: The Formulator’s Question Playbook

Spray drying is the leading method for producing amorphous solid dispersions (ASDs). It is also the process in which many preventable failures occur when transitioning from lab-scale powder to robust commercial manufacturing. Formulators frequently ask us the same questions: Which solvent? What feed load? What temperatures? What is the scale-up risk? What is the target for residual solvents?

We address these questions using a digital twin of the spray-drying process, which traces the droplet, solvent mixture, and API/polymer pair from the feed solution to the dry particle, including secondary drying. This Q&A compiles the 13 most common questions we receive when developing spray-dried ASDs. They are grouped into four categories: solvent selection, process conditions, secondary drying and scale-up.

Solvent Selection

1/ Which solvent should I use for spray drying my ASD?

The ideal solvent dissolves the API and polymer together at the highest joint concentration, prevents liquid-liquid phase separation (LLPS) during drying, and evaporates at realistic outlet temperatures. There is rarely a single answer. For any API/polymer pair, we provide a shortlist of candidates, ranked by where their maximum solubility sits relative to the LLPS boundary.

2/ How does amofor find the best solvent for my API and polymer?

We use PC-SAFT to model the interactions between the API, polymer, and solvent, including miscibility, solubility, and compatibility, as well as how these evolve when the solvent evaporates. The output is a ternary phase diagram of the API, polymer, and solvent showing the miscible region, the LLPS region, the glass-transition surface, and the solubility limits all in one image.

Next, we overlay the drying trajectory, which is the path a droplet traces through composition space as the solvent evaporates. A safe solvent remains in the single-phase region until vitrification, whereas a risky solvent crosses LLPS for any meaningful duration. For more details see Dohrn et al., 2021, Int. J. Pharmaceutics.

3/ How do I determine solvent ratio in solvent mixtures?

In binary or ternary solvent systems, the ratio determines the trajectory’s curvature. Even slight variations in solvent content can shift the trajectory from fully miscible to phase-separated, even with the same API/polymer pair. This occurs because one solvent evaporates faster than the other. Within milliseconds, the composition of the droplet shifts and the trajectory bends through regions of the phase diagram that the bulk feed solution would never reach. This is also why the solvent that dissolves everything cleanly at room temperature often performs worst during drying; it can still drive demixing as it leaves.

We map the trajectory across the full ratio range and identify the inflection point at which phase separation becomes thermodynamically inevitable. We always recommend a ratio with a safety margin beyond the inflection point rather than the ratio at the edge. For more details see Dohrn et al., 2021 Mol. Pharmaceutics.

Process Conditions

4/ What is the optimal solid feed load for spray drying?

The solid feed load, i.e., the mass fraction of API plus polymer in the feed solution, is one of the key factors that determines the trajectory of a droplet through the API/polymer/solvent phase diagram during drying. The optimal feed load is the highest concentration at which the droplet still has sufficient time to solidify into a glass before liquid–liquid phase separation or crystallization can occur. The phase diagram of the specific API, polymer, and solvent system defines this boundary. We therefore select the highest feed load that keeps the drying trajectory within the safe region.

In practice, higher and lower feed loads each have advantages and limitations. Higher feed loads increase process throughput because more solids are processed per unit volume of solvent, allowing larger quantities of material to be spray dried more efficiently. However, increasing the solids content also increases solution viscosity. At some point, viscosity can become the limiting factor for pumping, atomization, and stable process operation.

Lower feed loads, on the other hand, generally provide a larger safety margin for complete API dissolution and reduce viscosity-related processing challenges. The trade-off is lower productivity, since a larger amount of solvent must be evaporated to produce the same quantity of ASD, resulting in longer drying times and higher solvent consumption.

For most ASD development programs, we typically recommend feed loads in the single-digit percentage range. Depending on the API, polymer, and solvent system, substantially lower concentrations may sometimes be required to ensure complete dissolution and process robustness. In other cases, particularly with favorable polymer–solvent combinations, feed loads approaching 20 wt% can be achievable.

Unlike standard experimental screening approaches that often focus on a narrow concentration range, our modeling evaluates a broad spectrum of feed loads. We systematically assess the impact of formulation composition and process conditions across the entire practically relevant range and are not limited to conventional concentration windows. This allows us to identify opportunities for higher throughput while maintaining ASD quality and physical stability.

5/ What process temperatures should I set in the spray dryer, and how do I determine a safe process window?

Inlet temperature, outlet temperature, and feed rate are the primary operator-controlled factors. Together with atomization conditions and drying-gas flow, these parameters determine the coupled temperature-and-composition trajectory of the droplet from feed to dry particle. The outlet temperature is the key factor because it determines the residual solvent content and wet Tg at the operating point, i.e., the distance the particle is from its glass transition point when it lands in the cyclone.

The safe window is the range of outlet temperatures wide enough to produce consistent particles, yet narrow enough to prevent the failure modes noted in Question 6. The width of this window varies greatly between systems. Most CDMOs apply a default temperature envelope to very different systems, which is the single most common avoidable failure mode in early development. We provide a process design map that identifies the window for your system and explains where each edge lies.

6/ What can go wrong inside the droplet; and how do I tell which failure mode is which?

Most spray-drying failures remain undetected until characterization. There are four classic modes:

  • Wet particles: The droplet leaves the chamber before vitrifying, resulting in soft, sticky agglomerates instead of a free-flowing powder
  • Intra-particle demixing: The particles appear normal externally, but contain API-rich and polymer-rich domains
  • Residual solvent above the ICH limit is trapped in the glassy matrix and is slow to remove (see Question 9)
  • Cyclone deposits often lead to yield and cleanability losses, which are often misread as a hardware problem.

Each mode corresponds to a different region of the phase diagram. The SOLCALC software tells you which boundary your trajectory crossed rather than leaving you to diagnose the problem after a failed batch.

“Spray drying is one of the few operations where the entire fate of the product is decided in a window you cannot see and cannot sample. The digital twin allows us to look inside that window. The phase diagram is the map, and the drying trajectory is the route.Christian Lübbert, Ph.D.

7/ Why might a feed that appears miscible on the bench still fail in the dryer?

Because what happens in those milliseconds is not equilibrium. As the droplet shrinks, the solvent evaporates from the surface, forming a concentration gradient from the surface to the core. In a binary solvent system, the more volatile component leaves faster. This causes the droplet to curve through composition space instead of moving in a straight line. Even a small difference in volatility can bend that curve across regions of the phase diagram that the bulk feed would never reach at equilibrium. The bench miscibility test characterizes the starting point, and the trajectory determines the outcome.

Secondary Drying and Final Product Properties

8/ How long should secondary drying take; and how can it be reduced?

Even with high temperature and strong vacuum, most manufacturers still need two or more days because residual solvent kinetics are estimated empirically, batch by batch. We use a PC-SAFT model to compute the minimum residence time to reach the target residual at a given temperature and the maximum temperature the glass will tolerate. This model accounts for evolving Tg, the API’s interaction fingerprint, and solvent–polymer compatibility.

Where applicable, we employ acceleration strategies: a nitrogen sweep reduces the solvent’s partial pressure, while an assisting solvent (water or methanol) increases molecular mobility, enabling trapped molecules to escape. Across the campaigns we have modeled, secondary drying times have decreased by 50–75%; often dropping from days to less than one day; without compromising the final residual content or long-term stability.

9/ What residual solvent content can I expect in the final particles?

All spray-dried materials contain trace solvents. Typical as-collected levels are around 5% (50,000 ppm). The ICH Q3C(R9) guidelines set targets such as 600 ppm for dichloromethane and 5,000 ppm for ethanol. This process is slow for mechanistic reasons; many solvents have a large diameter and diffuse through the dense ASD matrix as if they were threads tangled in spaghetti.

PC-SAFT-based modeling can predict the residual solvent levels at the end of primary and secondary drying separately. More importantly, it can tell you in advance whether the planned step will reach the target level or if the solvent is structurally trapped. If the latter is true, the solution lies upstream: change the solvent system or feed composition before secondary drying can be effective.

10/ Will my final particles be glassy and stable, or will they recrystallize?

Every ASD is metastable. The relevant question is not whether it will crystallize, but rather, under what conditions the energetic barriers will hold. The phase diagram directly answers this question.

  • The miscibility window shows whether your composition is within a single-phase region or near a demixing boundary.
  • The wet Tg curve shows if the particle remains glassy at storage temperature after being plasticized by moisture.
  • The humidity overlay shows if ICH conditions (25 °C/60% relative humidity (RH) or 40 °C/75% RH) push the formulation into a phase-separated region.

If the target is in an unfavorable region, phase separation is thermodynamically inevitable, even if the initial stability data appear clean.

Benchmarked against more than 150 long-term stability datasets, our model predicts crystallization onset with an accuracy of ±20% (Grönninger et al., 2024, Molecular Pharmaceutics). Additionally, we have demonstrated that two ASDs with the same drug, polymer, and ratio can have a shelf life that differs by a factor of 1,000, ranging from six months to six thousand years, based solely on the cooling rate. ASDs have thermal memory; if the droplet does not relax into a dense glass, the kinetic stability collapses, even when the phase diagram appears favorable.

Scale-Up and Deliverables

11/ How do I predict scale-up risks before going to large scale?

Although lab and commercial spray dryers differ in terms of droplet size, residence time, and heat-and-mass-transfer geometry, the thermodynamics do not change with scale. The phase diagram remains unchanged; only the trajectory shifts. We compute the trajectory at the target scale and identify the specific risks: These include LLPS moving into the operating window, residual solvent climbing above the target level, the Tg landing at operating temperature, and sticky cyclone deposits.

The deliverable is a scale-up risk assessment available before the first technical batch, applicable to both early- and late-stage manufacturing.

12/ Can amofor model spray drying beyond ASDs?

Yes, our daily work involves substances beyond the Rule of Five, such as those with high molecular weight, high lipophilicity, and complex hydrogen-bonding networks. This framework also extends to adjacent unit operations that are governed by the same thermodynamics. Solvent casting, the slow cousin of spray drying and a technique used for mucoadhesive films and early ASD prototypes, has the same phase diagram and LLPS and vitrification boundaries. The only differences are in trajectory shape and residence time. Recently, we also assisted Bend Bioscience in modeling an original core-shell inhalation particle for lung cancer therapy (Anderson et al., 2025, Molecular Pharmaceutics).

13/ What deliverables do I get from an amofor spray drying simulation?

A standard amofor spray-drying simulation yields the following results:

(1) a ranked list of solvents and the mechanistic reason each candidate is included or excluded;

(2) a ternary API/polymer/solvent phase diagram with overlaid drying trajectories for each candidate solvent and feed load;

(3) a process design map showing the safe and risky regions and failure modes for each inlet, outlet, and feed rate combination;

(4) the predicted residual solvent content at the end of primary and secondary drying;

(5) the predicted wet glass transition temperature of the as-collected particle and the stability margin under ICH conditions;

(6) a scale-up risk register at the target manufacturing scale;

(7) the recommended secondary drying time and temperature, with the achievable reduction quantified against your current process.

Each deliverable is mechanistic, explaining the “why,” not just the “what.”

Talk to us about your spray drying project

If you’re designing a spray-drying process for an ASD, whether for candidate selection, troubleshooting, or scale-up, consult Dr. Christian Lübbert early on. Computing a phase diagram in week one can save six months of empirical screening that will never converge.

1/ What is the difference between higher and lower feed loads? 2/ what is the typical solids range amofor recommends for spray-drying ASDs? 3/ what feed loads does your modeling typically scan? 4/ is there a published amofor paper that quotes feed load?