Introduction
There is a wealth of published material on specific pharmaceutical applications of XRD [1-6]. The non-destructive nature and relative ease of sample preparation make XRD ideal here. In particular, the XRD pattern represents a crystalline drug “fingerprint” needed for patent descriptions, and to identify different drug batches. Other XRD pharmaceutical applications are excipient compatibility, optimization of process parameters, detection of form impurities, crystal morphology of active, and monitoring batch or dosage uniformity.
The purpose of this article is less specific, and directed more toward the non-expert XRD user; i.e., how XRD can be practically utilized to solve real-life problems in preformulation and formulation. In these situations, XRD is normally part of a multi technique approach. Therefore, we show XRD in context with some of these techniques. We also discuss disadvantages of XRD, and describe techniques potentially useful as alternative methods.
Mention is necessary regarding XRD sample preparation and data acquisition conditions. These aspects are not trivial and can lead to serious errors if improperly performed. Particle size, particle orientation, and data collection parameters influence data quality. The sample must be representative of the bulk material with particles uniformly distributed. The need for sample pre-treatment (grinding, etc.) should be assessed; if employed, samples phases must not be altered. Many methods are available for XRD sample preparation, each having advantages and disadvantages. These aspects are discussed by Buhrke et al [7].
Experimental
Samples were run on quartz zero background holders with a commercially available XRD system. A liquid nitrogen cooled solid-state Germanium detector was used with Cu k-alpha radiation at 45kv/40ma.
Case 1: Drug Substance Form Identification
Knowledge of drug substance form(s) is essential, especially identification of the most stable form. Polymorphism and pseudo-polymorphism (hydrated or solvated crystal forms) often are discovered during screening. Because physical changes in drug substance can potentially affect solubility, ease of manufacturing, bioavailability, and product stability, forms discovered in screening may require further solid-state characterization. Techniques like HPLC, solution NMR, and mass spectrometry are important for verifying chemical purity and composition, but not crystal form.
However, XRD in conjunction with other techniques can be quite informative. Consider Compound A, isolated as either a dihydrate or monohydrate. The amount of water in Compound A can be quantified by TGA, but TGA weight-loss curves for the hydrates do not give any information regarding physical structure. DSC thermograms of the two forms are similar, although the monohydrate has a higher melting point. While FTIR can be sensitive to O-H vibrational modes of hydrates, here the differences between the two forms were slight. But XRD—a non-destructive technique, unlike DSC—definitively distinguished the two forms. Note that as an alternative technique, solid-state NMR could also provide this information.
The relationship between the two hydrate forms was further examined using variable temperature XRD (VTXRD). Of interest was whether the dihydrate converted to the monohydrate upon heating. The VTXRD data, Figure 1, suggest no interconversion. The monohydrate appears more stable, retaining its crystal structure at 300oC, while the dihydrate crystal structure has disappeared. These observations agree with DSC and hot-stage microscopy data.
Characterization of the Compound A hydrates is relatively straightforward; application of XRD and TGA allow definitive identification of the monohydrate and dihydrate. Complications occurred when an intravenous formulation of Compound A was needed. The drug substance had a very limited solubility at neutral pH; to achieve the target drug concentration, the solution pH range was adjusted in buffers (pH 4 to pH 5). Several co-solvent combinations (e.g., with propylene glycol) were also evaluated. Initially all solution formulations were clear, but haziness developed, then precipitation. The precipitate was bright yellow, suggesting a transformation of the pale yellow dihydrate starting material. Precipitation appeared related to the amount of co-solvent, although it was also observed in buffers alone. To identify the precipitate, a characterization protocol was developed.
Precipitate material was isolated from bulk solutions by centrifugation or vacuum filtration, and air-dried; drying is necessary to remove co-solvent. Solution methods (HPLC, LC-MS, NMR) confirmed the precipitate was chemically identical to Compound A. The precipitate DSC, Figure 2, indicated it was a hydrate. The dehydration endotherm (Event 1) showed a Tmax similar to that of the known forms. The precipitate melting point (Event 2) was similar to that of the dihydrate. The precipitate exhibited an additional thermal event, a broad endotherm with Tmax at 70oC, suggestive of increased water.
Likewise, the precipitate FTIR spectrum exhibited modest differences in the O-H stretching region (3300-3500cm-1) compared to the other hydrate forms. So, both DSC and FTIR suggest the precipitate was a hydrate, but these results do not tell us anything about water stoichiometry or crystal structure. The former is confirmed by TGA, indicating the precipitate is a Compound A tetrahydrate. With respect to tetrahydrate crystal structure and its relationship to the other hydrate forms, XRD can address this question, Figure 3. The tetrahydrate clearly has a crystal structure distinct from the other hydrate forms.
It was later found that both monohydrate and dihydrate convert to tetrahydrate under high humidity conditions. The tetrahydrate has a lower solubility in the solution formulation, causing precipitation. What initiates tetrahydrate formation is not totally understood. Increasing the amount of co-solvent favors tetrahydrate appearance, but this would not account for observed precipitate in buffer solutions alone. Thus, solution pH may also play a role.
A potential challenge here was the relatively small amount of precipitate isolated from solution (frequently less than 5 mg), highlighting the need for sample conservation and avoidance of nondestructive characterization methods if possible. XRD data acquisition conditions were adjusted for sample size, Table 1.
Again, solid-state NMR could have been utilized in this work, although the smaller sample amounts (<5mg) might have been problematic.
Case 2: Stability of an Amorphous Drug Substance
Poor water solubility of drugs poses a difficult challenge for formulation. Many approaches have been taken to overcome this problem: Particle size reduction is often the first option pursued, but introduces the possibility of contamination from the grinding equipment. Another approach is to prepare the drug substance as an amorphous form. This can lead to a significant enhancement of solubility. However, it should be emphasized that amorphous formulations are inherently unstable and may crystallize, especially at high humidity storage. If amorphous material precipitates from a delivery system, drug dissolution and bioavailability may be compromised. Thus, it becomes critical to identify the onset of crystallization, and if possible set a detection limit for crystalline content.
The present example involves Compound B. Crystalline B, with or without PVP, was dissolved in methanol and isolated by rotary evaporation; spray-drying was carried out for scale-up. Table 2 shows that amorphous B has significantly enhanced solubility.
Amorphous B can be characterized by various techniques for information about crystallinity. In optical microscopy, crystalline and amorphous B appear dramatically different. Crystalline B exhibits obvious birefringence and well-defined particle morphology; amorphous B shows no birefringence and has much smaller, less distinct particles. While microscopy is an excellent tool for confirming amorphous material existence, it cannot make quantitative estimates of crystalline content in the amorphous matrix.
Another technique is modulated DSC (MDSC), which allows determination of the glass transition (Tg). For dry amorphous B, Tg is quite high, ca. 210oC, suggesting this material will remain stable if moisture is minimized. It is well known that increased moisture content lowers Tg, making the amorphous material increasingly susceptible to crystallization. This has significant implications for product storage; i.e., by minimizing moisture content, the amorphous form should remain stable. Tg is thus useful as a predictive tool, but can’t directly quantify crystalline content.
Regarding the onset of crystallization, XRD is useful to detect crystallinity in an amorphous matrix. This is because the crystalline and amorphous patterns are strikingly different: Crystalline forms exhibit relatively sharp, well-defined peaks, while amorphous forms display a diff use “halo.”
Preparing mixtures of crystalline and amorphous reference drug forms offers a way to quantify the amount of crystalline drug, Figure 4. The intensity or area of selected crystalline peaks diminishes as the amount of crystalline material in the amorphous matrix decreases; this change can be applied for quantification.
Typically XRD assesses crystallinity from peak height or area. Using the peak area at 10.4 2-theta gave the best linearity (Figure 5). The data also indicate the limit for detection of crystalline material is below 2.5%.
Note that PVP is amorphous, so its presence does not interfere with XRD crystallinity detection. However, it is not uncommon for crystalline excipient peaks to overlap with active peaks in drug product, necessitating selective subtraction procedures [8].
While application of XRD for crystallinity quantification can work well, there are caveats. One is preferred orientation, which will affect peak intensities. This can sometimes be overcome by grinding, or by using a capillary. With grinding, care must be taken that the sample does not change physically. Another problem relates to the quality of the reference standards and homogeneity of the mixtures used; incomplete mixing can lead to large errors when a small amount of one component is not uniformly distributed.
In addition, one must consider underlying assumptions of the quantification model. Here the two-state model was used, i.e., completely ordered (100% crystalline) and completely disordered (100% amorphous). However, this model may not truly describe the crystalline lattice disordering process [9]. Furthermore, there is concern that the “amorphous halo” concept is too simplistic for meaningful analysis. Treatment of more complicated amorphous systems is described in the literature [3,10].
While XRD and DSC are the most widely used techniques for determination of crystalline (or amorphous) content, many others are available. Shah et al. compares the merits of several methods [11].
Case 3 Identification of Active in Solid Dosage Form
Besides the “amorphous” strategy of Case 2, there are other ways to enhance solubility. One is salt formation; salts tend to be significantly more soluble, with a higher dissolution rate than the original acid or base drug substance. This strategy was applied to Compound C, a poorly soluble free acid. Here free acid starting material is treated with excess sodium hydroxide and a carrier (e.g., mannitol) acting as a conversion aid. This yields a crystalline dispersion, facilitating transformation of free acid to sodium salt [12]. The resulting in situ sodium salt requires characterization at different intervals of the formulation process, including the finished product (tablet).
Before salt form assessment can be done in such samples, one must consider excipients also present. In the mannitol formulation, the non-active components include: mannitol, microcrystalline cellulose, cornstarch, and a disintegrant. It is necessary to obtain the XRD pattern of each excipient to determine any areas of overlap with the drug substance. Furthermore, other conversion aids were evaluated for formulation; these were also analyzed by XRD. From the patterns of the excipients and various conversion aids (D-sorbitol, various non-ionic surfactants, propylene glycol, and PEG 400), the 2-12 2-theta region appeared best for sodium salt form determination. No excipients or conversion aids exhibited peaks there; all observed peaks were unique to the drug substance.
One concern relates to the identity of the sodium salt within the dosage form. Interestingly, an automated salt screen was unable to find any sodium salts of Compound C. Manual screening was more successful. Seven sodium salts were synthesized; all were crystalline, with differing XRD patterns. However, only salt forms 6 and 7—those containing the carrier material with sodium hydroxide—had XRD patterns matching those of the situ salts in the dosage form, Figure 6.
Initially it appeared that salt forms 6 (dihydrate) and 7 (trihydrate) were unrelated, but this turned out not to be the case; it became clear that forms 6 and 7 could interconvert rapidly, within one hour. Depending on the humidity level, one or both forms could be present, which is reflected in the XRD pattern.
Another concern is maintaining the desired product solubility. It is essential that the free acid be totally converted to the corresponding salt form(s) to avoid precipitation. XRD and FTIR are both useful for monitoring this. From Figure 6, it’s evident the main peaks of the free acid are generally distinct from those of the salt forms, suggesting a qualitative and possibly quantitative XRD method is feasible.
However, due to other considerations—availability of instrumentation at different manufacturing sites—an FTIR method was chosen. As with XRD, the FTIR method used known mixtures of acid and salt to establish a detection limit for the free acid, in this case by monitoring disappearance of the carboxylic acid carbonyl at 1734cm-1, where the sodium salt has minimal absorbance. A direct correlation between free acid content and carbonyl peak absorbance loss was established, with a limit of quantification (LOQ) of 5% and a limit of detection (LOD) of 2% acid.
Note with either XRD or FTIR, there was never any evidence of acid in the solid formulations, suggesting conversion to salt was complete.
Representative granulations were stored under different conditions and analyzed by XRD. Results are summarized in Table 3.
Despite the limited angular range, Forms 6 and 7 exhibit distinctly different patterns. Fortuitously the peaks at 11.3 2-theta (form 6) and 4.5 2-theta (form 7) are well separated and easily identified. Compound C granulations generally contain mixtures of Form 6 (dihydrate) and Form 7 (trihydrate). But the mannitol formulation, which contains only form 6 with desiccant, shows Form 7 formation without desiccant, as indicated in Table 3.
XRD works well in distinguishing forms 6 and 7, even in a tablet. FTIR cannot distinguish these two forms; on the other hand, parallel studies with Raman spectroscopy gave good results on the same samples, suggesting this technique could be an excellent back-up method for XRD.
Conclusion
The three examples given are intended to illustrate the diversity of XRD applications in pharmaceutical analysis. In Case 1, XRD was used to identify drug substance forms, including an unknown material. In Case 2, XRD was applied to quantify crystalline content in an amorphous formulation. In Case 3, XRD was used to identify two related drug forms in a solid formulation.
In summary, as a stand-alone technique XRD is very useful for problem solving in pharmaceutical analysis, although sample preparation and data interpretation require care. Also, by combining it with other techniques—FTIR/Raman, DSC/TGA, etc.—XRD can provide greater clarity and completeness in the understanding of events.
Acknowledgements
We thank Valerie Jobeck and Nathaniel Schuster for excellent technical contributions. We also thank Dr. Anthony Severdia for his encouragement and keen scientifi c insights.
References
1. Davidovich, M., Mueller, R., Raghavan, K., Ranadive, S., Vitez I., Sarsfield, B., The role of powder X-ray diffraction as a powerful tool in characterization of various hydrates of drug substance, American Pharmaceutical Review, 2001, 4, 53-60.
2. Sampath, I.S., Phadnis, N.V., Suryanarayanan, R., Quantitative analysis of complex pharmaceutical mixtures by the Rietveld method, Powder Diffraction, 2001, 16, 20-24.
3. Bergese, P., Colombo, I., Gervasoni, D., Depero, L.E. Assessment of the X—ray diffraction-absorption method for quantitative analysis of largely amorphous pharmaceutical composites, J. Applied Crystallography, 2003, 36, 74-79.
4. Harris, K.D.M., Modern applications of Powder X-ray diffraction in pharmaceutical sciences, American Pharmaceutical Review, 2004, 7, 86-91.
5. Davidovich, M., Gougoutas J.Z., Scaringe, R.P., Vitez, I., Yin, S., Detection of polymorphism by powder X-ray diffraction: Interference by preferred orientation American Pharmaceutical Review, 2004, 7, 10-17
6. Varasteh, M., Deng, Z., Hwang, H., Kim, Y.J., Wong, G.B., Quantitative determination of polymorphic impurity by X-ray powder diffractometry in an OROS formulation, Int. J. Pharm., 2009, 366, 74-81.
7. A Practical Guide for the Preparation of Specimens for X-ray Fluorescence and X-ray Diffraction Analysis, Buhrke, V.E., Jenkins, R. Smith, D.K. (eds.), Wiley VCH, 1998
8. Phadnis, N.V., Cavatur, R.K., Suryanarayanan, R., Identification of drugs in pharmaceutical dosage forms by X-ray powder diffractometry, J. Pharm Biomed. Analysis, 1997, 15, 929-943.
9. Rani, M., Govindarajan, R., Surana, R., Suryanarayanan, R., Structure in trehalose dihydrate— evaluation of the concept of partial crystallinity, Pharm. Res., 2006, 23, 2356-2367
10. Bates, S., Zograf,i G., Engers, D., Morris, K., Crowley, K., Newman, A., Analysis of amorphous and nanocrystalline solids from their X-ray diffraction patterns, Pharm. Res., 2006, 23, 2333-2349.
11. Shah, B., Kakumanu, V.K., Bansal, A.K., Bansal, J., Analytical techniques for quantification of amorphous/crystalline phases in pharmaceutical solids, J. Pharm. Sci., 2006, 95, 1641-1665.
12. United States patent 583771
Author Biographies
Cynthia Randall received her degrees in chemistry from the University of Wisconsin-Madison and the University of Michigan. Since joining the Analytical Sciences Department at sanofi-aventis, she has been primarily involved in solid-state characterization of new drug entities. Cynthia is the co-author of 17 scientific publications, including 4 book chapters. Her research interests include protein denaturation/stabilization, drug-liposome interactions, and pharmaceutical applications of spectroscopic and calorimetric techniques.
William Rocco was a Principal Research Investigator in the Pharmaceutical Sciences Department of sanofi-aventis for over 20 years. His primary interests are in preformulation and development of polymorphic forms. He has degrees in chemistry from SUNY-Oneonta and Chemical Engineering from SUNY Buffalo. He is currently a research chemist at Merck Research and Development.
Pierre Ricou received his doctoral degree from Ecole des Mines de Nantes (France), and then joined his now former employer sanofi-aventis as a postdoctoral scientist in the Analytical Sciences Department. Among his interests are the applications of XRD and XRF to characterize polymer formulations.