Engineering salt hygroscopicity through counterion choice — DVS gravimetric profiling, lattice-energy structure relationships, and a case study rescuing an unstable HCl salt
Roughly half of all marketed small-molecule drugs are administered as salts, and the hydrochloride is by far the most frequently chosen counterion — simple to form, cheap, and well precedented. But HCl salts are also disproportionately hygroscopic: the small, highly mobile chloride anion supports weak, low-density crystal packing that leaves interstitial space for water molecules to intrude. When a salt absorbs enough atmospheric moisture to dissolve part of its own crystal lattice (deliquescence), every downstream manufacturing and stability assumption breaks down.
Moisture uptake is not a cosmetic defect — it cascades through every unit operation:
Handling and processing: • Powder caking and loss of flow (Carr index rises from 18 to >35, "poor" flow) blocks hopper discharge during tableting • Electrostatic charging changes with adsorbed water layer, causing content non-uniformity in low-dose blends • High-shear wet granulation becomes unpredictable: an already-moist API adds uncontrolled liquid to the granulation endpoint • Milling and micronization: moist powders agglomerate on impact surfaces, reducing yield and broadening the particle size distribution
Physical stability during storage: • Deliquescence point (critical RH, RH0) for this HCl salt: 58% at 25°C — comfortably exceeded in tropical Zone IVb warehouses (ICH climatic zones) without dehumidification • Above RH0, the crystal surface dissolves in its own sorbed water film, forming a saturated micro-solution that can recrystallize as a different (often less soluble) polymorph or hydrate on drying • Caking in blister pockets and bottles: cohesive arch formation raises disintegration time from 4 min to >25 min after 3 months at accelerated conditions
Chemical stability: • Sorbed water acts as a reaction medium and proton donor for hydrolysis-sensitive functional groups (esters, amides, carbamates) • Degradation kinetics frequently follow pseudo-first-order dependence on %RH above a threshold (typically 40–60% RH) because bulk-like mobile water only appears once monolayer coverage is exceeded • Observed here: 2.1%/month total degradants at 40°C/75%RH vs. 0.15%/month at 40°C/dry (desiccated) — a 14-fold acceleration attributable almost entirely to sorbed moisture
Regulatory and CMC consequence: • ICH Q6A decision trees push a hygroscopic drug substance toward tighter packaging specifications (foil-foil blister, desiccant canister, controlled-humidity manufacturing suites — typically <30% RH) • Each added control is a recurring cost and a supply-chain fragility; sponsors increasingly treat "redesign the salt" as cheaper than "engineer around the salt" if caught before Phase 3 • Rule of thumb (Serajuddin, Adv. Drug Deliv. Rev. 2007): if the free-form salt-screen leader shows >15% w/w uptake at 80% RH, escalate to a broader counterion or cocrystal screen before committing to formulation development
Salt selection is now standard practice to run in parallel across a panel of pharmaceutically acceptable acids (for basic APIs) or bases (for acidic APIs), rather than defaulting to the first crystalline hit. High-throughput crystallization platforms let a CMC group generate and triage 10–20 candidate salts in the time it once took to characterize one.
Counterion shortlist for a basic API (pKaH ≈ 8.9), ranked by typical hygroscopicity precedent:
Strong mineral/sulfonic acids (robust salt formation, but often hygroscopic): • Hydrochloride (pKa −7): ubiquitous, cheap, but small ion → weak lattice, frequently hygroscopic • Sulfate (pKa −3, 1.99): higher lattice energy than HCl, moderate hygroscopicity, but stoichiometry (1:1 vs 1:2 hemi/di-sulfate) can be ambiguous • Mesylate (pKa −1.9): historically favored (Marks 1980s), though genotoxic alkyl sulfonate ester impurity risk (EMA 2019 mesylate guidance) now requires tight process control
Arylsulfonic acids (bulky, hydrophobic anions — often lower hygroscopicity): • Besylate (benzenesulfonic acid, pKa −2.8): large planar hydrophobic anion, dense herringbone packing, historically low hygroscopicity (amlodipine besylate precedent) • Tosylate (p-toluenesulfonic acid, pKa −2.8): similar packing advantages, additional methyl group increases hydrophobic surface area • Napadisylate (naphthalene-1,5-disulfonic acid, pKa −2.9, −2.9): divalent, forms particularly dense 2:1 salts; used for rescue of hygroscopic HCl programs
Carboxylic acids (weaker acids, moderate ΔpKa, generally more hygroscopic than arylsulfonates): • Maleate (pKa1 1.9): compact, high solubility, moderate hygroscopicity, some concern over Michael-addition reactivity with primary amines • Fumarate (pKa1 3.0): the trans isomer of maleate; substantially less reactive, moderate lattice energy, good developability precedent • Succinate, citrate: lower ΔpKa (<3) risk partial proton transfer — some batches show a spectrum of stoichiometries (0.5, 1, 1.5 equiv) rather than one crystalline phase
Triage funnel applied here: 1. Solubility >1 mg/mL in ≥3 crystallization solvents (ethanol, IPA, acetone, MEK, EtOAc, MeCN, water, THF) — enables practical process crystallization 2. Stoichiometric, single-phase crystalline hit by PXRD (no halos, no peak splitting suggesting solvate mixtures) 3. Confirmed 1:1 (or defined n:m) stoichiometry by ¹H-NMR integration and ion chromatography 4. Melting point / decomposition onset by DSC >150°C (manufacturability, avoids eutectic-driven caking) → Of 14 counterions screened, 9 produced usable stoichiometric crystalline salts; these advance to DVS profiling in Stage 3.
DVS is a gravimetric technique: a microbalance suspends a few milligrams of sample in a flowing gas stream of precisely controlled relative humidity, recording mass continuously as %RH is stepped from 0% up to 90–95% and back down to 0%. Because it measures true equilibrium mass change rather than a single-point snapshot, DVS is the reference method cited in ICH Q6A hygroscopicity classification and virtually every published salt-selection study.
DVS instrument mechanics (e.g. Surface Measurement Systems DVS Advantage, TA Instruments Q5000 SA):
• Ultra-microbalance resolution: 0.1 µg, enabling detection of sub-0.1% w/w mass changes • Gas delivery: dry N2 and water-vapor-saturated N2 mixed by mass-flow controllers to the target %RH, ±0.5% RH accuracy • Equilibrium criterion: dm/dt < 0.002%/min sustained for 10 min, or a maximum dwell time cap (e.g. 360 min) per step • Output: isotherm plot of % mass change vs. %RH for both sorption (increasing RH) and desorption (decreasing RH) branches
Classification schemes applied to the mass-change-at-80%RH value:
Callahan et al. (1982) scale: • Non-hygroscopic: <0.2% w/w increase at 80% RH • Slightly hygroscopic: 0.2–2% w/w • Moderately hygroscopic: 2–15% w/w • Very hygroscopic: >15% w/w, or deliquescence observed below 90% RH
Ph. Eur. 5.2.1 (European Pharmacopoeia) — near-identical four-tier scheme referencing mass increase at 25°C/80% RH after equilibration, now the default acceptance-criteria language in most European CMC dossiers.
Interpreting the isotherm shape (beyond the single 80% RH number): • Type II (sigmoidal) isotherm: monolayer coverage at low RH transitions to multilayer/bulk-like water above a threshold RH — typical of moderately hygroscopic crystalline salts • Sharp discontinuity (step change) in the sorption branch: signals a hydrate phase transition — the crystal is converting to a higher hydrate form at that RH, not merely adsorbing surface water • Large hysteresis between sorption and desorption branches: indicates the water is not fully reversible — often a sign of an amorphous fraction or a metastable hydrate that does not fully dehydrate on the down-ramp • Vertical mass runaway with no re-plateau: deliquescence — the solid is dissolving into its own sorbed water; the RH at which this begins is the critical relative humidity (RH0), directly measurable by DVS-deliquescence mode or predicted from the Grover/ICH RH0 correlation to solubility and activity coefficient
Applied to the 9 salt candidates here: HCl (18.4% @75%RH, RH0=58%, "very hygroscopic/deliquescent") anchors the poor end; besylate (0.6% @75%RH, RH0=88%) and napadisylate (0.3%, RH0=92%) anchor the "non-hygroscopic" end, spanning a >60-fold range in equilibrium water uptake across chemically reasonable counterion choices for the same free base.
Hygroscopicity is not a random property assigned to a molecule; it is a direct consequence of crystal packing energetics. A salt with a higher lattice energy holds its ions more tightly together, leaving less thermodynamic incentive for water molecules to insert themselves into the structure. Rational counterion selection therefore reduces, in large part, to choosing anions that maximize lattice stabilization while retaining acceptable solubility and processability.
Estimating lattice energy without single-crystal structures — the Hess-cycle (Born–Fajans–Haber) approach:
ΔH_lattice = ΔH_solution − ΔH_solvation(cation) − ΔH_solvation(anion)
• ΔH_solution measured directly by solution calorimetry (dissolving a weighed salt sample in excess water/solvent, recording heat evolved or absorbed) • ΔH_solvation terms estimated from ionic radius via the Born equation or taken from tabulated single-ion solvation enthalpies • This series: ΔH_lattice ranges from 612 kJ/mol (HCl salt) to 754 kJ/mol (napadisylate salt) — a >140 kJ/mol spread that log-linearly correlates with DVS uptake at 75% RH (R²=0.87 for this 7-point series)
Why bulky arylsulfonate/arylcarboxylate anions pack more densely:
1. Charge delocalization reduces local electric field: • Chloride is a small, hard, point-charge anion — its intense localized field is an excellent hydrogen-bond acceptor for water • Sulfonate charge is delocalized over three oxygens on a bulky aromatic scaffold; the effective field per unit surface area is lower, weakening the anion's affinity for individual water molecules
2. Hydrophobic surface area competes with hydrophilic sites for the crystal surface: • Besylate and tosylate present a phenyl/tolyl face that packs via π-stacking and CH···π contacts against the API's own aromatic rings — a "hydrophobic clamp" that excludes water from the interface • Napadisylate's divalent naphthalene core links two cations per anion, increasing crystal connectivity and coordination number
3. Channel and void volume: • PXRD-indexed unit cells for the HCl salt show an 8% larger unit-cell void volume (Platon SQUEEZE-estimated solvent-accessible volume) than the besylate salt — literally more empty space for water molecules to occupy • The besylate salt's hydrogen-bond network (N–H···O=S at 2.87 Å, confirmed by single-crystal XRD) fully satisfies the cation's H-bond donor sites internally, leaving no unsatisfied acceptor/donor sites exposed at the crystal surface to recruit atmospheric water
4. Channel hydrates as a failure mode to screen out: • Some low-symmetry salts form channel hydrates (water threading through 1-D channels in the lattice) rather than being truly anhydrous-stable; these show stepwise isotherms with hysteresis rather than smooth Type II curves and must be flagged even if the 80%RH uptake number looks acceptable • Slurry/competitive maturation experiments (API + 2 solvate/hydrate forms slurried together) confirm which crystal form is thermodynamically stable at the intended storage condition
Practical design rule distilled from this and analogous programs (cf. Morris et al., Int. J. Pharm. 2001; Stahl & Wermuth, Handbook of Pharmaceutical Salts, 2011): prioritize bulky, hydrophobic, delocalized-charge anions with an internally satisfied hydrogen-bond network when the API structure allows it — but always verify with DVS, since packing predictions from lattice-energy proxies alone have exceptions.
This program entered CMC development with a hydrochloride salt that failed accelerated stability: caking, deliquescence at ordinary warehouse humidity, and hydrolytic degradation exceeding the 1%/month ICH Q1A action limit. Rather than engineering elaborate packaging controls around a fundamentally unstable material, the team re-ran the counterion screen, selected besylate based on DVS and lattice-energy data, and re-established the entire CMC package on the new salt form within one development cycle.
Timeline and decision points for the counterion switch:
Month 0 — trigger: 3-month accelerated stability (40°C/75%RH) pull on the HCl salt tablets shows total degradants at 2.4% (spec NMT 1.0%) and visible blister-pocket caking; root-cause investigation implicates moisture ingress via the hygroscopic drug substance, not the excipients or process.
Month 1 — re-screen: existing 9-candidate salt-screen data (Stage 2) reviewed against DVS results (Stage 3) and lattice-energy ranking (Stage 4). Besylate selected over napadisylate (marginally lower uptake) because besylate's aqueous solubility (14 mg/mL at pH 1, biorelevant gastric pH) comfortably exceeds the BCS-relevant target, whereas napadisylate's bulkier divalent anion reduced free-API solubility by ~40% — illustrating that the "best" DVS number is not automatically the right process choice.
Month 2–3 — process development and scale-up: • Salt-formation crystallization optimized: API free base + besylic acid (1.02 equiv) in IPA/water (9:1), anti-solvent (MTBE) addition, seeded cooling crystallization • Kilo-lab scale-up (2 kg batches): isolated yield 92%, chemical purity >99.5% HPLC, residual solvent within ICH Q3C limits • Particle engineering: wet-milling to D90 <40 µm for dissolution-rate matching to the original HCl-salt clinical formulation, confirmed by f2 similarity factor (f2=68) in comparative dissolution
Month 4–6 — re-validation: • Full DVS re-confirmation on 3 independent GMP lots: 0.5–0.7% uptake at 75%RH, RH0 83–89% — consistent with screening-scale data, confirming the property is intrinsic to the salt form and not an artifact of small-scale crystallization • Forced degradation and 6-month accelerated stability (40°C/75%RH, open dish and packaged): total degradants 0.24% at 6 months packaged vs. projected >8% for the HCl salt under the same protocol • High-shear wet granulation trial: no capping or lamination observed at commercial-scale mixer torque, versus intermittent capping previously seen with the HCl salt at high fill-weight batches (attributed to localized moisture plasticization) • Simplified control strategy filed: standard PVC/PVdC blister with foil lid, no desiccant canister required; manufacturing suite humidity control relaxed from <30% RH to standard <60% RH ambient
Economic and regulatory impact: • Packaging cost reduced ~40% (desiccant canister and secondary aluminum pouch eliminated) • Manufacturing suite dehumidification capital and operating cost avoided • Shelf-life claim supported to 24 months at 25°C/60%RH long-term storage (ICH Q1A Zone II), versus a 12-month claim with cold-chain-adjacent storage recommended for the HCl salt • Single amendment to the CMC module (new drug substance specification, updated stability data package) rather than a full reformulation program, because dose, bioavailability, and clinical PK were preserved by matched dissolution
The single largest lever in this entire program was not a formulation excipient or a packaging tier — it was the counterion. Switching from hydrochloride to besylate reduced equilibrium water uptake more than 30-fold, raised the critical relative humidity by 30 percentage points, and cut hydrolytic degradation by over 50-fold, at a cost of roughly six months of salt-form re-screening. The general lesson for CMC teams: run a broad counterion screen with DVS characterization before Phase 2b, when switching form is still cheap — not after a stability failure forces the switch under time pressure.