First-order hydrolysis: can the oral solution reach a 24-month shelf life?
A pharmaceutical team is developing a 50 mg mL⁻¹ oral solution of a drug that hydrolyses by first-order kinetics. An accelerated study at 60 °C gives 92.0% of the label claim remaining after 30 days. A second study at 70 °C gives 89.0% remaining after 30 days. The team proposes a 24-month shelf life at 25 °C with a specification of not less than 90.0% remaining. The activation energy is estimated from the two temperatures using the Arrhenius equation. The formulation is buffered at pH 3.0, where the observed rate constant follows k_obs = k_H[H⁺], with k_H = 0.68 L mol⁻¹ day⁻¹ at 25 °C; a pH-independent water reaction is negligible over the pH range studied. A reviewer questions whether the 60 °C data alone are sufficient and whether the proposed shelf life is realistic.
1. Calculate k at 60 °C and 70 °C, estimate the activation energy, and predict k and t90 at 25 °C. Then calculate the percentage remaining after 24 months at 25 °C and decide whether the specification is met.
Read the worked answer
For first order, k = −ln(C/C0)/t.
At 60 °C: k60 = −ln(0.920)/30 = 0.0834/30 = 2.78 × 10⁻³ day⁻¹
At 70 °C: k70 = −ln(0.890)/30 = 0.1165/30 = 3.88 × 10⁻³ day⁻¹
Arrhenius: ln(k70/k60) = (Ea/R)(1/T60 − 1/T70)
ln(3.88/2.78) = 0.335
1/333.15 − 1/343.15 = 8.75 × 10⁻⁵ K⁻¹
Ea = 0.335 × 8.314/(8.75 × 10⁻⁵) = 3.18 × 10⁴ J mol⁻¹ = 31.8 kJ mol⁻¹
Extrapolate to 25 °C:
k25 = 2.78 × 10⁻³ exp[−(3.18 × 10⁴/8.314)(1/298.15 − 1/333.15)]
= 2.78 × 10⁻³ exp(−1.35) = 7.2 × 10⁻⁴ day⁻¹
t90 = ln(10/9)/k25 = 0.1054/(7.2 × 10⁻⁴) = 146 days ≈ 4.8 months
After 730 days: C/C0 = exp(−7.2 × 10⁻⁴ × 730) = exp(−0.526) = 0.591, i.e. 59.1% remaining.
The specification of not less than 90.0% at 24 months is therefore not met. The formulation is far too unstable at pH 3.0; the predicted t90 is only about 5 months.
2. The team wants to reformulate by changing the buffer pH. Using the given acid-catalysis rate law, calculate the maximum k25 allowed for 90.0% at 24 months and the pH needed to reach it. Decide whether pH 4.0 gives an adequate margin.
Read the worked answer
Required: C/C0 ≥ 0.900 at t = 730 days.
kmax = −ln(0.900)/730 = 0.1054/730 = 1.44 × 10⁻⁴ day⁻¹
Current k25 = 7.2 × 10⁻⁴ day⁻¹, so the rate must fall by a factor of 7.2 × 10⁻⁴/1.44 × 10⁻⁴ ≈ 5.0.
With k_obs = k_H[H⁺] and k_H = 0.68 L mol⁻¹ day⁻¹:
[H⁺] = 1.44 × 10⁻⁴/0.68 = 2.12 × 10⁻⁴ mol L⁻¹
pH = −log(2.12 × 10⁻⁴) = 3.67
So pH ≈ 3.7 is the threshold. At pH 4.0, [H⁺] = 1.0 × 10⁻⁴ and k = 6.8 × 10⁻⁵ day⁻¹, giving t90 = 0.1054/6.8 × 10⁻⁵ ≈ 1,550 days, about 4.2 years. That is a comfortable kinetic margin over 24 months.
However, the decision is not automatic. Moving from pH 3.0 to 4.0 may reduce solubility, bring the drug close to its pKa, or allow a base-catalysed pathway that was negligible at pH 3. The reformulation must be checked experimentally with a pH-rate profile, and the buffer capacity and excipient compatibility at pH 4.0 must be confirmed.
3. The reviewer asks whether the 60 °C data alone would have been sufficient, and what else is needed before accepting the Arrhenius prediction. Give a reasoned response.
Read the worked answer
No. A single temperature gives only one rate constant and cannot separate k from Ea, so it cannot support an extrapolation to 25 °C. Even the two temperatures used here give only a two-point Arrhenius line, with one time point at each temperature and no replicate. Two points can hide curvature, analytical error or a change in mechanism.
A defensible package would include:
- at least three temperatures, each with several time points, to confirm first-order behaviour from linear ln(C) versus t plots;
- replicate vials and a validated assay with known precision, because a 3% change over 30 days is small;
- a check that the Arrhenius plot is linear over the studied range, since a change in the rate-determining step would make extrapolation unsafe;
- a pH-rate profile around pH 3–6 to test the assumed acid-only rate law;
- stress studies for oxygen, light, metal ions and container interactions.
The current data are useful for a preliminary ranking, but they are not sufficient to lock a 24-month shelf life without confirmation at or near the intended storage temperature.