The ability to quantify moisture in real-time is critical for optimising fluid bed drying operations. Chris Heil, of Thermo Fisher Scientific, looks at the growing use of FT-NIR for rapid in-process moisture determination in dryers
Drying in the solid oral drug industry is a central step in the manufacture of tablets and granulates, which make up 80% of the drugs in the marketplace. Drying is the science of evaporating water or an organic solvent using heat. Different temperatures and times are required depending on the product's quality and the desired residual moisture. As the evaporation of the solvent cools the product, relatively high temperatures can be used for drying without overheating the product. At the same time, it is important to mix the product well during drying.
Fluid bed dryers are effective for solid drying in the pharmaceutical industry. The heat transfer is excellent and the drying time is ideal for gentle drying of the product. The whole
surface of each individual particle is available for drying due to the fluidisation of product in the dryer. A homo-geneous product temperature and a uniform drying of the starting product are achieved as a result of the continuous and thorough mixing in the dryer. This cannot always be guaranteed in the case of contact and radiation dryers.
Batch fluid bed dryers are used whenever reproducible product quality and comprehensive process documentation are required. They provide a high level of economy with gentle product handling. The drying process starts with moist product being fed into the dryer. The heated air stream coming from the bottom of the dryer mixes the product vigorously. The product is suspended as it is mixed and dried by the heated air stream. The optimum amount of heated air blown into the dryer depends on the particle size and density of the product. The temperature of the air can be adjusted during the process. Once the product is dry, it can be cooled before the dryer is emptied. This process dries the product to the desired moisture content efficiently.
Continuous fluid bed dryers are used whenever high throughput is required. The steps that are carried out for each batch in the above process are carried out in different places in the continuous process. Moist starting product is fed in continuously and moves through the drying system. Effective drying is accomplished using different temperatures of the incoming air for the different chambers. If the last chamber is operated with cold air, the dry product can be cooled before leaving the drying system. The batch fluid bed drying systems are the most popular in pharmaceutical manufacturing. There are three different types of dryers based on what direction the heated air is blown into the dryer: top spray, bottom spray and tangential spray.
Near Infrared (NIR) analysis is an accepted technique in the pharma industry (European Pharmacopeia (Ph.Eur.) 2.2.40 1997, ASTM NIR Method E 1944, USP Chapter <1119> and the Japanese Pharmacopeia). It is non-destructive and provides real-time data; it is finding widespread use for in-process analysis of moisture in multiple industries. Fourier Transform (FT)-NIR is an excellent tool for implementing Process Analytical Technology (PAT) in the drying process.
PAT means incorporating science into manufacturing by conducting analysis in real-time to control the process. Currently, pharmaceutical companies rely on recipe-based manufacturing (e.g. heat for 2 hrs, mix for 25 mins, etc.). This does not lead to any kind of process understanding so entire batches can simply be thrown out without understanding the root cause of failure. Processes are still tethered to one-off QC lab testing that determines the fate of an entire batch. PAT aims to work process understanding into the manufacturing process, mitigating risk and improving the bottom line.
FT-NIR spectroscopy can play a large role in PAT since it can meet the requirements across the product development lifecycle from r&d to manufacturing. Many FT-NIR analysers use a common platform so that transfer of methods between instruments from lab to process can be achieved without a decrease in method performance. In addition to in-line dryer analysis, FT-NIR has been used for other process analyses including raw material ID, blend uniformity, tablet content uniformity, and API quantification.
FT-NIR has many advantages over traditional Karl Fischer and Loss on Drying methods for moisture analysis (Table 1) including increasing dryer operational efficiency and lowering operating cost without sacrificing accuracy or precision.
The challenge in any moisture determination, whether in-line or in the lab, is getting a representative sample to the instrument. Fluid bed dryers use the action of hot gas and fluidisation to dry product as it swirls around inside the dryer. This swirling action results in a variable distance and density of the product in front of a measurement probe. This negative factor can be overcome by chemometric processing and the averaging of large amounts of spectral data.
Another issue is whether the product will form a crust or leave a residue at the probe-sampling interface. Instruments that analyse using sight glasses on the side of a dryer cannot overcome this obstacle since there is no way to clean the glass while the dryer is running. Also, the material that is adhered to the sight glass is not representative of the product that is in the dryer. This issue can be overcome with a purgeable tip or retractable probe with a self contained cleaning system. The purgeable tip probe (Thermo Scientific) (Figure 2) removes powder or sample crust using pressurised N2 blown through the holes on the end of the probe.
The retractable probe (Figure 3) can be cleaned automatically by solvents in a separate self-contained chamber. The purgeable tip probe works best for product under 15% moisture since it can be easily blown off the end of the tip. For product that cannot be removed from the purgeable tip probe (higher moisture product), the retractable probe is the best choice. The retractable probe can be mechanically cleaned in between each measurement and then solvent cleaned and dried in between each dryer run. The movement, purging and data collection of these two probes can be controlled automatically using Thermo Scientific RESULT software with the process controller.
For this study, a diffuse reflectance fibre optic probe with purgeable tip was installed into a fluid bed dryer (see Figure 1). The dryer ran in a cycle of 120 seconds of purge drying followed by 10 seconds of filter shaking. Spectra were collected during the 10 seconds of filter shaking because of better reproducibility of results than collecting spectra during the purge step. This allowed a new data point to be produced every two minutes.
A Thermo Scientific Antaris MX FT-NIR analyser was used for the spectral collection. The instrument acquired 12 scans per sample at 8 cm-1 resolution from 4800-10000 cm-1. The primary method analysis was by Karl Fischer with each sample being run in duplicate, which took about 10 minutes. A large set of dryer batches were sampled with the samples ranging from 1-23% moisture.
The method developed for moisture determination used the Partial Least Squares (PLS) algorithm with five factors. The spectra were pre-processed before method development using 1st derivative, Multiplicative Scatter Correction (MSC) and a Savitzky-Golay filter with a data point of seven and polynomial order of three for smoothing. The MSC processing helped to account for differences in effective pathlength caused by the varying distance of the sample from the probe and also the effects of light scattering due to the different particle size of the samples. The 1st derivative helped peak resolution and also removed unwanted baseline features.
The spectral region used in the method was 4850-9980 cm-1. This region incorporates the combination band for water at 5150 cm-1 and the 1st overtone for water at 7000 cm-1. As can be seen in Figures 4 and 5, the region around 7000 cm-1 shows significant variation in absorbance due to the large range in moisture (1-23%).
Quantifying water in most compounds is straightforward by NIR. This is because water is the most sensitive compound that can be measured by NIR. Most compounds can be detected reliably down to 0.05-0.1%, while water sensitivity is roughly an order of magnitude better. The method for moisture determination produced a RMSEC (root mean square error of calibration) of 0.56 with a correlation coefficient of 0.995 (see Figure 6). The RMSECV (root mean square error of cross validation) for the method was 1.05 with a correlation coefficient of 0.981.
In conclusion, an FT-NIR analyser equipped with a purgeable tip fibre optic reflection probe can accurately measure moisture in a fluid bed dryer. The fit of the calibration method (R = 0.995) has proven to be excellent across a wide range in moisture (1-23%). The speed at which real-time moisture analysis data is available using FT-NIR makes determining the dryer endpoint much easier than thief sampling for primary wet chemistry analysis. The dryer endpoint can be determined based on when the moisture concentration hits a control value. The need for lab equipment, solvents and personnel to constantly analyse dryer samples is eliminated. The accuracy, precision, linearity, and speed of analysis by FT-NIR allows for a definitive determination of dryer endpoint.
Knowing exactly when a dryer has reached its endpoint will save energy, eliminate the destruction of product due to over-drying, and increase the overall efficiency of the drying process.