| Tablets accounted for up to 80%of the total market share of pharmaceutical dosage forms are one of the most widely used dosage forms.However,the mechanism of particle compressed into tablet is still lack of systematic knowledge that properties of tablets are evaluated via hardness,fractability and so on according to the pharmacopoeia.The discrete element method that can analyze the movement behavior of individual particles in particle flow will provides convenience for elucidating the forming mechanism of tablets.In this paper,starch,an elastic excipient,taken as the model,was investigated combining physical and numerical experiment.The discrete element parameters of starch was predicted by BP neural network,and the reasons for the difference of different measuring methods of Angle of respose were analyzed.Then the evolution rule of meso-structure during particle densification and the formation mechanism of tablets was further demostrated.This"white box"theory is helpful to improve us understand of tablet quality and provide data reference for intelligent drug production in the future.The main results are as following:(1)The structure,morphology and compressibility of starch were characterized by FT-IR,XRD,SEM,Angle of repose and powder direct pressing.The results showed that the functional groups and crystal types of different types of starch had no obvious difference.The shape of starch particles was spherical,irregular block or column,and the average particle size of starch particles was less than 50μm except for direct-pressed pre-gelatinized starch and soluble starch.The Angle of repose obtained by different measurement methods is different in size and above 30°.Among the starch types studied,only the pre-gelatinized starch can be compressed,but under the same pressure,different types of pre-gelatinized starch can be pressed into plain tablets with different hardness,fragility and disintegration time.(2)Three methods of Angle of rerest measurement(lifting cylinder method,shear box method and funnel injection method)were used to construct training sample data,and BP neural network was used to find the sample data for training and testing.Discrete element parameters of starch were predicted according to the relationship,and verified.The results show that when the number of neurons in the hidden layer of BP neural network is 11,the coefficient of determination R~2of training sample and test sample is 0.9999 and 0.9409respectively,and the predicted output can reach the expected output.The change law of particle coordination number in different measurement methods is extracted and the internal factors of the difference of Angle of repose measurement are analyzed from the meso-perspective.(3)The starch particles with different shapes were quantified by aspect ratio(AR),and the discrete element simulation of compaction of starch particles with different shapes was established by using the calibration parameters,and the densification mechanism was analyzed from a mesoscopic perspective.The results show that particles with AR=0.5 are more prone to splinter phenomenon after compaction,which is caused by uneven density distribution of particles with AR=0.5 after compaction.Cluster particles were established by particle bonding method,and discrete element simulation was carried out based on EEPA model for particle compression process and hardness test process.Through mesoscopic analysis,it was found that the formation of force chain was related to the pore structure and contact force of starch particles during the compression process and hardness test process. |
PDF Full Text Request