Kinetics of Phase Transitions in Amorphous Carbamazepine: From Sub- Tg Structural Relaxation to High-Temperature Decomposition

Abstract

Thermokinetic characterization of amorphous carbamazepine was performed utilizing non-isothermal differential scanning calorimetry (DSC) and thermogravimetry (TGA). Structural relaxation of the amorphous matrix was described in terms of the Tool-Narayanaswamy-Moynihan model with the following parameters: Δh* ≈ 200-300 kJ·mol^-1, β = 0.57, x = 0.44. The crystallization of the amorphous phase was modeled using complex Šesták-Berggren kinetics, which incorporates temperature-dependent activation energy and degree of autocatalysis. The activation energy of the crystal growth was determined to be >320 kJ·mol^-1 at the glass transition temperature (T_g). Owing to such a high value, the amorphous carbamazepine is stable at T_g, allowing for extensive processing of the amorphous phase (e.g., self-healing of the quench-induced mechanical defects or internal stress). A discussion was conducted regarding the converse relation between the activation energies of relaxation and crystal growth, which is possibly responsible for the absence of sub-T_g crystal growth modes. The high-temperature thermal decomposition of carbamazepine proceeds via multistep kinetics, identically in both an inert and an oxidizing atmosphere. A complex reaction mechanism, consisting of a series of consecutive and competing reactions, was proposed to explain the second decomposition step, which exhibited a temporary mass increase. Whereas a negligible degree of carbamazepine degradation was predicted for the temperature characteristic of the pharmaceutical hot-melt extrusion (~150 °C), the degradation risk during the pharmaceutical 3D printing was calculated to be considerably higher (1-2% mass loss at temperatures 190-200 °C).

Keywords: carbamazepine; crystal growth; structural relaxation; thermal decomposition.

Figure 1

Raw DSC heating curves (right column); crystallization signals with subtracted thermokinetic background (left column). Exothermic effects evolve in an upward direction.

Figure 2

Temperature programs for the CR and CHR structural relaxation experiments (left column), and the corresponding DSC responses (right column). Exothermic effects evolve in an upward direction.

Figure 3

Selected joint DTA-TGA signals obtained for the CBZ decomposition in N₂ and air atmospheres (left column)—exothermic effects evolve in a downward direction; the (right column) depicts full sets of TGA measurements.

Figure 4

The top left and right micrographs show the amorphous CBZ sample heated to 80 °C for 5 and 10 min, respectively. Bottom graphs show the Raman spectra (full and zoomed-in on the low-phonon part) of the amorphous, as-purchased, DSC crystalized, and surface-crystallized (crystals from the micrograph above) CBZ samples.

Figure 5

(Top graph): Evaluation of Δh* from CR and CHR cycles with the listed values corresponding to the linear fits (bottom and left axes); Δh*-T dependence determined from the CR cycles evaluation fit by the 2nd-order polynomial (top and right axes). (Middle graph): Normalized set of CHR cycles. (Bottom graph): Application of the simulation-comparative method to the present data—points show the experimental data from the middle graph, thin lines depict selected simulated dependences for various β&x combinations, thick red line shows the best correspondence between the experimental and simulated data (β = 0.57, x = 0.44).

Figure 6

(Top left graph): Kissinger evaluation of E using the temperatures corresponding to the crystallization peak onset and maximum with the listed values corresponding to the linear fits (bottom and left axes); E-T dependence determined from the T_p values fit by the 2nd-order polynomial (top and right axes). (Top right and middle graphs): q⁺ dependences of kinetic parameters determined by means of the sc-MKA method. (Bottom graphs): Example crystallization signal fit by either one or two AC processes.

Figure 7

(Top left graph): Kissinger evaluation of E using the temperatures corresponding to the TGA points of inflection determined from the derivative TGA data for the two step-like mass losses (bottom and left axes); E-T dependences determined from the Kissinger dependences fit by the 2nd-order polynomials (top and right axes). (Top right graph): Example TGA data (measured under N₂ atmosphere) fit by the sc-MKA method employing the depicted reaction scheme. (Bottom graphs): Kinetic predictions calculated using the kinetic parameters determined for the 0.1 °C·min⁻¹ TGA data-curve obtained under N₂ atmosphere—the predictions were calculated for isothermal annealing at denoted temperatures characteristic for the processes of pharmaceutical hot melt extrusion and 3D printing.

Figure 8

(Top graph): Crystal growth rate data taken from [25] and transformed to evaluate the activation energy E_G. (Bottom graph): Activation energies determined for various thermally induced processes occurring in (amorphous) CBZ. “Crystallization” and “crystal growth” denote the data from DSC and microscopy, respectively.

Scheme 1

Molecular structure of carbamazepine.