Spectroscopy-based Excipient Composition Analysis

Molecular spectroscopy plays an important role in the physical characterization of pharmaceutical excipients solids. Based on the theory of infrared and Raman spectroscopy, BOC Sciences is able to accurately interpret the spectral data of the most commonly used excipients in pharmaceutical dosage formulations by identifying the observed spectral features.

Our Capabilities

Analysis of the composition of pharmaceutical excipients often involves a range of spectroscopic techniques to determine the molecular and chemical composition. Only by using all these techniques can a reliable determination of the composition be accomplished.

Raman Spectroscopy

In recent years, Raman technology has gained a noticeable market in the identification of raw materials for pharmaceuticals. Compared to traditional analytical techniques such as HPLC and NIR spectroscopy, Raman analysis is faster and cheaper. In addition, Raman techniques do not require sample preparation, direct sample contact with the sample and the ability to test samples through transparent packaging materials such as glass or plastic, making it an ideal tool for rapid identification of drug excipients. It is critical and required for pharmaceutical manufacturers to develop quality control procedures in place to ensure that drug excipients are correct and meet adequate quality standards.

Figure 1. Low-wave number Raman (LWR) spectroscopy. (Bērzi, K.; et al. 2020)

BOC Sciences has introduced advanced Raman spectroscopy as an effective and efficient technique for excipients identification, process analysis and final product certification. Our Raman spectroscopy is capable of producing compound-specific "signature" Raman peaks that are used to generate reliable chemical identifications. As a result, sample interference can be minimized while maximizing efficiency. Our Raman spectroscopy can be used to test the full range of materials covered by a 785 nm system, with the additional ability to cover materials blocked by fluorescence, thus providing the most comprehensive range of excipients identification.

• Clustering models obtained using principal component analysis (PCA)

At BOC Sciences, PCA is performed to classify the Raman spectra for several excipients. The spectra of the drug, excipients and APIs are run together in the PCA model and the variance described by each principal component is examined.

• Deep learning (DL) for excipient identification

We use DL-based component identification to solve the component identification problem in the Raman spectra of mixtures.

• Non-negative least squares (NNLS) for excipient quantification

Raman spectral analysis introduced by BOC Sciences can provide sufficient spectral resolution for the identification of drugs. Raman spectra are information rich but not easy to interpret, especially for spectra affected by auto-fluorescence phenomenon, and the signals of high-dose pharmaceutical excipients may mask the signals of low-dosed APIs, especially those with weak Raman responses. The identification results of the pharmaceutical excipient components can be further refined by the non-negative least squares method, and the ratios of each component can also be accurately calculated. The search results can be further refined by using non-negative least squares method and the proportion of each excipient used in the drug can be estimated.

Figure 2. Raman mapping technology. (Shuoyang, Z.; et al. 2019)

Near-infrared (NIR) Spectroscopy

For the production of pharmaceutical products, all ingredients must be reliably identified. Near-infrared reflectance spectroscopy speeds up the identification of excipients. Based on NIR spectroscopy, we are able to achieve a 100% classification of pharmaceutical ingredients and excipients. Our team has used NIR spectroscopy to collect spectral data of different types of pharmaceutical excipients. For each class, at least 15 samples of excipients are collected for the database establishment, taking full account of the different batches and suppliers. We can build models by Bayesian algorithm, support vector machine algorithm, and K-nearest neighbor algorithm paired with first-order difference, second-order difference, MSC, and SNV preprocessing.

Figure 3. Pharmaceutical Raw Material Identification with Near-Infrared Spectroscopy. (Hui, Z.; et al. 2020）