In-Process Analysis of Moisture in a Fluid Bed Dryer Using Near-Infrared Spectroscopy

Robert Mattes, FOSS NIRSystems, Inc.,

Abstract In-line moisture measurements of a pharmaceutical granulation of lactose, microcrystalline cellulose and crospovidone were performed in a fluid bed granulator/dryer using top sprayed granulating liquid. A near-infrared (NIR) prediction model was developed for moisture on spectra collected during a calibration run. Subsequent granulations were analyzed for moisture content real-time throughout the granulation and drying process using the NIR process instrument. NIR spectroscopy fits in well with the Process Analytical Technology (PAT) initiative as developed by the FDA. Introduction Near-infrared spectroscopy (NIRS) is an analytical technique based on absorption measured in the near-infrared region of the electromagnetic spectrum that is between the visible and the mid-infrared. The fundamental absorption bands of functional groups occur in the mid-infrared and are very strong. The overtone absorptions of these fundamental bands occur in the near infrared (NIR) spectral region and allow direct measurement without sample preparation due to the relative weakness of absorption. The OH, CH, NH and SH bonds have the strongest overtone absorbance in the NIR region. Samples are typically withdrawn with a thief during processing for moisture analysis and analyzed off-line in a laboratory for moisture content. Commonly there is a delay before analysis results are available to the operator and processing decisions are made without optimal product moisture information. End point is often based on elapsed time or product temperature and not moisture content. One of the elements of the PAT initiative is to use in-line analysis to increase process understanding and control and to verify product quality and release it for subsequent processing without delay. Using NIRS, the process can be monitored for low levels of residual moisture and alcohols for better process control and end-point determination. Methods The NIR instrument used to collect spectra was the FOSS XDS Process Analyzer (FOSS NIRSystems, Laurel, MD). See Figure 1. Spectra were collected in the reflectance mode from 800 nm to 2200 nm with 0.5 nm data intervals and 32 scans were co-added to produce a single spectrum. Spectra are shown from 800 nm to 2100 nm where NIR absorbance is strongest. The XDS Process Analyzer spectra are useful out to 2200 nm where a roll-off due to throughput loss in the fiber optics occurs. A Spoon probe of novel design for the fluid bed application was installed into a Niro MP 2/3 Precision Granulator at a 45 angle to the central axis of the product container. Note the collection spoon and purge vents located on the probe tip. See Figure 5. After each NIR spectrum the software sends an electronic signal that activates an air purge clearing the spoon for a new sample. A 12 kg. charge of lactose (Pharmatose 200M, DMV), microcrystalline cellulose (Avicel PH 101, FMC) and crospovidone (Polyplasdone XL 10, ISP) was prepared by Niro and loaded into the product container . The product was fluidized for five minutes to blend the dry mixture to homogeneity. An aqueous solution of 15% Polyvinyl pyrrolidone (Plasdone K29/32) was top sprayed at 1.5 Bar pressure. NIR spectra were collected every 40 seconds. Samples for loss on drying (LOD) analysis were withdrawn with the thief every five minutes. A 2.0 gram sample was analyzed for LOD at 160 C for 15 minutes using the Mettler Toledo HR73 Halogen Moisture Analyzer. When the water/binder solution pump was stopped the drying process began. The drying operation was uniform and gradual over a period of 15 minutes. Discussion Figure 7 shows the raw non-math- pretreated NIR spectra of dryer scanned inprocess. Notice the spectral baseline shift as the scattering increases due to particle agglomeration. Figure 8 shows the same spectra treated with the second derivative to remove the baseline shift and to enhance spectral absorbance features due to moisture. Figure 9 is an enlargement of the analytical moisture band where the O-H stretch/bend combination occurs. Due to the second derivative treatment, the moisture increases downward in this region. Partial least squares (PLS) regression was used to develop the prediction model which uses principal component analysis and is a variation of principle component regression (PCR). A two factor PLS regression model was developed using the second derivative intensity over the range 900 - 2100 nm with an R2 value of 0.9896 and a standard error of calibration (SEC) of 0.2171. Figure 10 shows the NIR predicted moisture level versus the LOD values as predicted by the model. Validation samples taken during subsequent granulation runs were predicted for moisture. The standard error of prediction (SEP) was 0.4232 %. See figure 11. The PLS prediction model was used to predict on spectra taken throughout the whole process from beginning to end. Figure 12 is a process trend chart showing the moisture level during the process from blending to spraying the binder/moisture solution and drying.

Figure 8. Second derivative of dryer spectra scanned in-process.

Figure 9. An analytical region where moisture absorbs due to the O-H stretch/bend combination band.

Figure 10. Calibration set for the PLS model, R2 = 0.9896, SEC = 0.2171% moisture..

Figure 11. Validation samples taken on later subsequent granulation runs. SEP = 0.4232 %.

Figure 12. A process trend chart showing the process from blending to spraying the binder/moisture solution and drying. This is a typical FOSS Vision routine analysis output trend chart. Conclusion The FOSS XDS Process Analyzer demonstrated the ability to predict samples of lactose and Avicel with residual amounts of water during the fluid bed granulation/drying operation. This trial demonstrates the ability of the FOSS Fluid Bed Probe (Scoop Probe) to measure a fluidized sample for residual moisture. Other residual granulating liquids and constituent levels could be measured using the same instrument/probe configuration. The correct NIR probe must be placed in the product container in a manner that provides sufficient sample contact with the probe tip window. Correct probe design and proper placement in process equipment is of high importance for success of NIR implementations. Endpoint determination can be made when the moisture level asymptotically approaches a lower limit during the drying cycle or the percent change reaches a minimum. References Figure 7. Raw NIR spectra of product in dryer scanned during processing. Notice the spectral baseline shift as the scattering increases due to particle agglomeration. This effect was minimized by applying the second derivative. R. A. Mattes, D. E. Root, A. P. Birkmire, In-line Process Analysis of Residual Moisture in a Fluid Bed Granulator-Dryer using Near-Infrared Spectroscopy (Spectroscopy, Jan. 2005) Acknowledgments Andrew P. Birkmire the process engineer at Niro Pharma Systems, Inc., Columbia, MD who prepared the charge and operated the granulator.

Figure 1. FOSS XDS Process instrument with standard fiber optic probe.

Figure 2. Niro MP 2/3 Precision Granulator .

Figure 3. The product container opened.

Figure 4. Inside the product container with the NIR probe at lower left.

Figure 5. NIR Spoon probe of novel design.

Figure 6. NIR probe with product on the Spoon.