Monitoring continuous mixing process dynamics through the NIR spectra baseline using a stream sampler

The uncontrolled powder flow through the interface used for process monitoring of direct compression continuous manufacturing results in baseline variation of the near infrared spectra. This baseline variation is then eliminated through spectral preprocessing. In this study the baseline was controlled with a stream sampler and used to evaluate process dynamics during continuous mixing. The stream sampler was operated at a paddle speed of 8 RPM to maintain a throughput of 35 kg/h while ensuring uniform filling of all paddle segments. The Moving Block Standard Deviation (MBSD) method was used to monitor the real-time baseline variation and was integrated within the synTQ program. A threshold baseline MBSD was determined at mass steady state of five calibration blends (40-60 % w/w acetaminophen). The spectra acquired from powder flowing through the stream sampler exhibited a reproducible and stable baseline during mass steady state, across different days and setups. The stable baseline was attributed to the flat surface achieved by the confined powder flow within the stream sampler. The threshold baseline MBSD successfully distinguished mass steady-state and mass flow variation during continuous mixing runs at 35 kg/h throughput. Partial least squares regression models were investigated with and without spectral transformation. The model without spectral transformation presented lower prediction bias and root mean square error of prediction for the independent test blends. Variographic analysis was useful to understand the overall process variance including sampling and analytical errors.