However, the new Medical Cannabis industry needs to adopt the quality standards of the pharmaceutical industry and at the same time satisfy the specific requirements of Cannabis-legislation. This means that the Cannabis products for medicinal use need to comply with its established quality parameters and standards and to be manufactured under cGMP regulations. Tetrahydrocannabinol  and cannabidiol  are well known phytocannabinoids, considered as the most notable Cannabis components with pharmacological activity. THC acts mostly as a CB1 receptor agonist, leading to its distinguished psychoactive and pain relief effects, while CBD works through a variety of pharmacological pathways, including inhibition of endocannabinoid reuptake, activation of transient receptor potential vanilloid 1 and G protein-coupled receptor 55. Considering their immense importance in the overall pharmacological activity, as well as the current Cannabis legislation restrictions, related to the THC content of some classes of Cannabis products, one could entitle the content of CBD and THC as critical quality parameters of Cannabis products. The CBD and THC content could also be considered as critical material parameters when Cannabis flowers are used as starting material in the production. The quality control of Cannabis inflorescence and chemotype differentiation is a subject of many Pharmacopoeias focus on the assay of the content of five main cannabinoids CBDA, CBD, CBN, D9 THC and D9 -THCA. The above-referenced Pharmacopoeias comprise monographs on cannabis defining the dried drug or herbal substance ‘‘Cannabis inflorescence” that contains a minimum of 90 and a maximum of 110% the amounts of cannabinoids such as D9 -THC and CBD, as well as Cannabinoid carboxylic acids such as D9 -THCA and CBDA, calculated as D9 -THC or CBD, based on the dried drug. Depending on the content of D9 -THC and CBD, authorities have classified generally three chemotypes of Cannabis sativa L.: D9 -THC predominant type, i.e. drug-type , CBDpredominant type, i.e. fiber type and intermediate chemotype .

Regarding cannabis extracts, so far the German pharmacopeia recognizes discontinued cannabis extract – Cannabis extractum normatum defining as an extract from whole or shredded, flowering, dried shoot tips of the female plants of Cannabis sativa L. that contains D9 -THC at least 1% and at most 25%  for the extract, and CBD maximum 10%  for the extract. The chemical complexity of cannabis makes its pharmaceutical standardization challenging. Thus, vertical grow rack chemical characterization must include well-defifined methodologies that would characterize the plant chemotype and the herbal drug as well as extraction procedures. It was found that the concentrations of target cannabinoids obtained for the same plant chemotype originating from different suppliers could vary for more than 25%. scientific and technological development in regard to Cannabis sativa began, highlighting the need of sensitive, specific and robust analytical methods for identification and quantification of the active constituents of Cannabis sativa. Gas chromatography  and liquid chromatography  are regarded as analytical methods of choice for the quantification of phytocannabinoids. Both are relatively slow and costly techniques and require sample preparation that involves at least extraction of the ingredients with organic solvents. Since quality assessment in the current Medical Cannabis industry relies on end-product testing, these techniques are vastly employed. However, the variability of phytocannabinoids content in the plant material often exerts an issue in the inconsistency of the finished product quality parameters. Sampling problems and sample representativeness is a major limitation in the end-point testing, particularly when the expected variation of the product quality parameters is high. Usually, physical limitations  prevent the test sample analysis to adequately represent the whole batch variability. Besides, the critical quality parameters are not monitored frequently during the production process, thus lacking the knowledge for appropriate identification of critical process parameters in the process optimization. Therefore, there is an obvious need for the introduction of the concept of quality by design  in Cannabis products manufacturing, which means that product quality should be scientifically designed to meet specific objectives, not merely empirically derived from the performance of test batches. In this manner, the product and process characteristics important to desired performance must be derived from a combination of prior knowledge and vast experimental assessment during product development. The generated data will result in the construction of a multivariate model linking product and process measurements and desired parameters.

Implementation of Process Analytical Technology  is an important QbD tool aimed to increase the understanding essential for the quality throughout the manufacturing process, which became recognizable in the Pharmaceutical industry by the initiative launched by the FDA. Infrared spectroscopy is a promising analytical technique that is consistent with the PAT requirements of the FDA guidelines, and its implementation depends on the advances in instrumentation and chemometrics that will facilitate the qualitative and quantitative aspects of the technique. In the literature, few attempts have been made to introduce near-infrared spectroscopy for quantification of phytocannabinoids in plant material, in liquid pharma-grade Cannabis formulations and for growth-staging of Cannabis. To the best of our knowledge, so far no scientific papers are addressing the capability of mid-infrared spectroscopy  for quantification of phytocannabinoids in Cannabis plant material and extract. Therefore, our present work aims in highlighting the potential of mid-infrared  spectroscopy as PAT in the quantification of the main phytocannabinoids , considered as critical quality/material parameters in the production of Cannabis plant and extract. To achieve our goals, we have performed MIR analysis of Cannabis flowers  and Cannabis extracts from various randomized sources with various CBD and THC content and employed a multivariate statistical approach to develop and optimize the calibration models. Furthermore, a prediction set was used to estimate the prediction capability of each MIR model against the referent analytical technique .MIR spectra were collected on the attenuated total reflection module of an Alpha Platinum ATR Fourier transform infrared spectrometer . Ten milligrams of each dry sample was used for spectral acquisition whereas few drops of each extract were placed on the ATR plate compartment. After the collection of each extract IR spectrum, the solidified specimen that remained on the plate was dissolved in hexane and the compartment was cleaned to proceed with the next extract sample. Each IR spectrum was recorded in the 4000 to 400 cm 1 region and averaged from eight scans with the spectral resolution adjusted to 4 cm 1.The partial least-squares analysis was employed to build calibration models for quantification of THC and CBD in Cannabis extract and flowers, using the spectroscopy skin of Simca14 . Spectral pre-processing was performed using the spectral filters add-in. The correlation coefficients of both X and Y matrices , the predictivity coefficient , root mean square error of estimation , and the root mean square error of cross-validation  were used as the main statistical indicators. The VIP and coefficient plots were used for further analysis of the models.

The predictive capability of the models was evaluated on separate predictive sets and the root mean square error of prediction  was determined for each model.The description and assignment of the IR spectra was conducted following the available literature with an emphasis of the analytical bands responsible for differentiation of the cannabinoids of our interest: THCA, THC, CBDA, and CBD . Firstly, given the need for precise spectral examination, two air-dried, non-thermally treated flowers  were screened . The strikingly various content of CBDA and THCA in these flowers enabled to ascribe the bands originating from these cannabinoids because it was assumed that the remaining content of the chemical species of the flower is rather similar and comparable . Namely, the attempt was made to select discriminating spectral regions for THCA and CBDA native flowers that, although depict evident similarities, exhibit marked differences . The apparent spectral disparity was discussed in terms of:  shift of the position of the dominant bands,  absence/presence of discriminating bands and  differences in the intensity of the identical band . The applied approach was advantageous because the cannabis flowers contain various molecule species and was perplexing to specify an infrared band solely to a particular cannabinoid or terpene analyte. Thus, the bands in the dominant THCA flower emerge at 1120, 909, and 711 cm 1 that are not registered in the CBDA flower where, on contrary, the bands at 3402 and 620 cm 1 evolved . An intensity increase of the 888 cm 1 band in the CBDA plant was observed, largely diminishing in the spectrum of the THCA flower . Another apparent difference among both spectra is associated with a marked shift of the medium to strong pair-analogues around 1570, 1250, and 1180 cm 1 . However, due to the similar structural formula of the tracked cannabinoids, we found all these regions as reliable indicators to fortify the discrimination of the THCA and CBDA molecular entities, but we will firstly focus on the interpretation towards the absent/present bands and further encompass the assignment in the appointed shifting band regions. The band at 3402 cm 1 found in the CBDA dominant flowers is attributed to the stretching OH vibration from the aromatic OH group situated in para-position to the carboxylic group fragment . This OH group is not present in the THC and THCA molecules  and therefore lacks in the spectrum of the FL1 cannabis cultivar . On contrary, the bands at 1120 and 909 cm 1 as well as the band at 711 cm 1 in the spectrum of THCA flowers  could be related to the aryl-alkyl ether group present in the THCA molecules  and assigned to antisymmetric and symmetric C–O–C stretching, and ring symmetric C–O–C bending vibration, respectively. On the other hand, the @CH2 wagging vibrations from the  gave rise to the strong 888 cm 1 band registered in the CBDA flower spectrum because the THCA compound lacks unsaturated C@C linkage in this structural position . Further focus on the assignment in the appointed shifting band regions infers that the 1570 cm 1 maximum is attributed to the stretching m vibration within the aromatic rings in the lignin, cellulose and the remaining cannabis grow racks species. C–H bending vibrations from the CH2 and CH3 aliphatic groups in these compounds are observed around 1430 cm 1 , and the m vibration next to the carboxylic group around 1180 cm 1 .

However, the most intense absorption around 1250 cm 1 , slightly shifted among the THCA and CBDA spectra, is ascribed to the m vibration from the carboxylic group since upon decarboxylation the intensity of this band majorly decreases . Although the precise assignment of these bands poses a great challenge due to the complex matrix of the samples, the proposed tentative assignment  was made in accordance with the limited IR spectral literature data for hemp and cannabis related specimens and the recent non-destructive Raman studies aimed to differentiate between hemp and cannabis.It is worth emphasizing that the decarboxylation protocol was delivered for non-origin correlated THC and CBD matrices of 45 thermally-treated flowers and 34 extracts. The idea was driven to develop a proper protocol that can sufficiently and reliably estimate the content of the major CBD and THC in any randomized series of flower and extract samples. For ease of interpretation, we have also considered the fact that the content of the acidic cannabinoids THCA and CBDA in the non-thermally treated samples is reasonably higher compared to the thermally treated flowers. Namely, flowers dominantly contain the acidic forms of the cannabinoids which upon careful temperature action are decarboxylated and converted to THC and CBD, respectively. The results from the IR spectra filtered the most important regions that reflect the major differences appearing among the different fresh flowers, and other regions where the major differences occur among the decarboxylated flowers and extract samples in comparison to the spectra of the starting fresh flowers. These spectral outcomes were found complementary to the chemometric results . To the best of our knowledge, the literature lacks IR band assignments for pure THCA, CBDA, THC and CBD, though the spectral appearance of these compounds in the native form can be found in the German Pharmacopeia, Specac App Note, Bruker App note, and PerkinElmer App note. The IR spectra of the acidic forms of the cannabinoids of interest  were only presented  by Hazekamp et al. who isolated these compounds from the concentrated ethanolic solutions , subsequently evaporating the ethanol under vacuum.