Introduction
Typically, the desired solid state for active pharmaceutical ingredients (drug substance) is crystalline. Amorphous content within the crystalline drug substance may have a deleterious impact on the drug product such as reduced chemical and physical stability.
Powder X-ray diffraction (PXRD) is one of several significant tools used to characterize the solid state character of drug substances. Randall et al [1] discussed the various uses of PXRD to pharmaceutical development as applied both to drug substance and drug product.
Powder X-ray diffraction is considered the ‘gold standard’ for identifying and quantifying crystalline phases in materials as diverse as minerals to active pharmaceutical ingredients. For pharmaceutical samples with amorphous content within the range of approximately 10 to 90% w/w, powder X-ray diffraction (PXRD) provides a means to determine the relative amorphous content. While less-routine XRD techniques may provide a lower limit of detection/quantitation for amorphous content, in the author’s experience, 10% amorphous content is a practical lower limit for detection using typical PXRD equipment.
The purpose of this article is to briefly mention the various PXRD approaches to quantify amorphous content. The majority of this article will focus on a simple approach that does not require standards and is suitable for routine pharmaceutical development.
Experimental
The data presented were obtained with a typical Bragg-Brentano powder diffractometer using a copper X-ray source and a nickel filter at the detector to provide diffraction data using λ= 1.54 Å .
Discussion
During the development of a drug substance synthesis, initial material may be completely amorphous or poorly crystalline. PXRD can be used to screen for the relative crystallinity of samples. At some point, it may be advantageous to determine quantitatively the level of amorphous content in the samples.
Quantitative phase analysis by PXRD is not trivial [2]. However, the complexities of analyzing a material such as a geologic sample arelessened for organic samples such as pure drug substance or drug product due to minimal sample X-ray absorption affects.
Figure 1-PXRD scan of sample X obtained using a back-filled cavity mount and a variable divergent slit. Scan collected in reflection mode. The sharp features (peaks) are due to the crystalline component. The broad hump in the baseline is due to the amorphous component. Background scattering unrelated to the amorphous and crystalline components produces an offset of the baseline from 0 counts.
Figure 1 shows the PXRD pattern for a mixture of crystalline and amorphous drug substance X. In a mixture of amorphous and crystalline material, the PXRD will exhibit both sharp and broad features. The sharp peaks are due to the crystalline component and the broad features (sometimes referred to as “halo”) to the amorphous component. Deconvolution of the mixture PXRD into separate PXRDs with sharp-only diffraction peaks and broad-only diffraction peaks will allow for determining the percentage of amorphous content.
Prior to applying quantitative methods to PXRD data for amorphous determination, the analyst must be sure that the broad features are due to amorphous (or glassy) material rather than due to microcrystalline or nanocrystalline material. The diffraction peak shape is affected by a variety of parameters, including the size of crystallites making up drug substance particles. As the crystallite size is reduced, the diffraction peaks broaden. Once the size is sufficiently reduced, the crystalline diffraction peaks may have broadened to the extent to merge into each other forming a single broad diffraction peak (halo). If a single-crystal X-ray diffraction data is available, the powder diffraction pattern can be calculated with different peak shapes and widths to simulate the PXRD of micro/ nanocrystalline material. Such material would be crystalline but less ordered due to the reduced repetition of molecular packing in 3-dimensional space compared to material exhibiting PXRD with sharp diffraction peaks. Analysis for microcrystalline material is a separate topic not discussed here. In the absence of single-crystal XRD data, the analyst should discuss with the sample submitter the history of the sample to understand if assaying for true amorphous content is appropriate. The presence of a glass transition in the differential scanning calorimetry profile of the material would serve as proof of the presence of amorphous material.
At the early stage and even later stages of development, it would be advantageous to use a method for determining amorphous content that did not require the use of standards. This approach is discussed next.
Approach 1 – PXRD Quantitation of Amorphous without the Use of Standards
For samples where little material (tens of mg) is available, silicon or quartz zero background plates may be used. Typically samples are dusted on the plates previously coated with a thin film of petroleum jelly or other viscous material. The level of the petroleum jelly for PXRD sample preparation is typically kept low to minimize the signal from the petroleum jelly which might be confused for amorphous drug substance. For the analysis of amorphous content, the use of petroleum jelly should be avoided. Thus, the sample should remain horizontal during the diffraction scan which is a feature of theta/theta diffractometers. Theta/two-theta diffractometers require the sample to tilt with the scan which may cause the powder to fall off the stage. Sample spinning to improve counting statistics makes the problem worse so the use of theta/theta systems for zero background plate preparations is recommended.
Where more material is available (200 mg or greater), the sample material may be filled into a sample cavity. For quantitation of mixtures of crystalline phases, back-filling [3] is recommended to avoid the effects of preferred orientation of particles when the sample is filled from the top of the cavity and smoothed flat with a glass slide. For the quantitation of amorphous/crystalline mixtures using the complete diffraction pattern, preferred orientation issues are less important. Thus front-filling of the sample is an option if back-filling is not available or problematic. A comparison of the reproducibility of various sample preparations on the raw and processed diffraction data is recommended.
Modern diffractometers allow for the data collection to be obtained with either fixed or variable divergent slits. The slit situated in front of the X-ray source, limit the divergence of the X-ray beam.
PXRD results collected using a fixed divergent slit is also known as “constantvolume” data. Starting from low two-theta and scanning to higher two-theta, the area illuminated by the X-ray beam gets smaller and the depth of penetration increases. Provided the area illuminated with X-rays does not exceed the sample boundaries and the sample depth is sufficiently thick that the X-ray beam will not penetrate to the bottom of the sample at the highest scanned diffraction angle, the sample volume interacting with the X-ray beam remains constant. PXRD results collected using a variable divergent slit is also known as “constant-area” data. The divergent slit is decreased as the scan goes to higher diffraction angles which results in a constant area of sample illumination but different sample penetration by the X-rays with varying diffraction angle. For a thin zero background plate preparation, the X-ray beam will completely penetrate the sample such that both a constant sample area and a constant sample volume are analyzed.
For cavity-amounted samples, the fixed divergent slit operation may be preferred so that a constant volume is analyzed throughout the diffraction scan. For zero background plate preparations, the variable divergent slit approach is recommended. In practice, data can be collected both ways during method development.
Where sufficient sample is available, a cavity-filled sample preparation is recommended to maximize the amorphous signal relative to the background.
In addition to the typical reflection geometry, modern research grade diffractometers allow for the option of collecting the diffraction data in a transmission mode. This alternate mode may be considered during method development.
For routine PXRD, the typical diffraction scan range is around approximately 2 to 45 degrees two-theta (with copper K-alpha radiation). Most PXRD vendor software allows the user to deconvolute the PXRD pattern to isolate the crystalline component. A blank scan should be run to determine the background artifacts from the diffractometer. Depending on the diffractometer optics, the lower diffraction limit for integration of the diffraction signal from the amorphous and crystalline components will be approximately 3 to 4 degrees two-theta (to avoid scatter from the direct X-ray beam at low angles). The upper scan limit should be approximately 70 degrees to allow proper estimate of the PXRD baseline.
Figure 2- PXRD scan of sample X following deconvolution of the diffraction pattern into the crystalline (portion above green trace) and amorphous (portion below green trace) components.
Figure 2 shows the resulting deconvolution of the sample PXRD scan presented in Figure 1. Different vendor software allow the user to adjust the resolution to discriminate between the gradual baseline changes due to amorphous content and background scatter. By subtracting out the portion of the data above the baseline that separates the crystalline from amorphous components, the diffraction intensity for the amorphous component can be determined. The ratio of the diffraction intensity for the amorphous component divided by the total diffraction intensity multiplied by 100 provides the approximation of the percent weight/weight amorphous content. The starting point (low two-theta) for integration is just above the point where scattering from the direct beam is observed. The ending point is where the diffraction signal returns to the background level.
The approach may result in background signal due to X-ray scattering by air and the sample holder being incorporated into the diffraction area associated with the amorphous component. The deconvolution process can be refined by running blank scans. For samples prepared using zero background plates, the blank scan would simply use a clean zero background plate with no applied sample. For samples prepared using cavity mounted sample holders, an empty sample holder is not appropriate since scattering from inside the sample holder may be observed. If available, a zero background plate inserted into the cavity can serve as a blank. The height of zero background plate should be flush with the top edge of the cavity. Applying a small piece of modeling clay affixed to the underneath side of the plate and then turning the sample holder over and pressing against a smooth and clean surface (such as as a flat bench top or a glass slide) can help properly align the plate with the cavity top. Inspection of the blank scans would allow for estimating the instrumental background and then subtracting this from sample scan prior to deconvolution of the sample diffraction pattern into the crystalline and amorphous components. The PXRD scan presented in Figure 2 had the instrumental background removed prior to deconvolution into amorphous and crystalline components. The lower the percent amorphous, the more attention to detail in background removal and integration parameters is required.
If possible, a sample containing a sizeable fraction of amorphous content (30 to 70%) should be analyzed by 13C solid-state nuclear magnetic resonance (SSNMR). SSNMR is a “nuclei-counting” technique and does not require external standards to quantitate amorphous content. The results from the SSNMR can be used to check the PXRD results and aid in optimizing the PXRD processing parameters, the choice of sample preparation, the choice of variable or fixed divergent slits, and the choice of reflection or transmission data collection. The SSNMR results are most trustworthy when there is a clear separation in the resonance amorphous and crystalline peaks. If there is overlap requiring deconvolution to separate the amorphous and crystalline components, then there is added uncertainty in the percent amorphous calculation. If the overlap in the SSNMR data is severe, then SSNMR may not be useful in providing an orthogonal check of the PXRD method.
Vibrational spectroscopy (IR and Raman) can also be used to check the PXRD method. However, the IR and Raman methods require external calibration curves that would involve preparing mixtures of pure and amorphous standards. Lack of homogeneous mixing, particle size/morphology differences between the calibration curves and actual samples, and differences in the nature of the amorphous standard and amorphous content in the sample will lead to errors in the estimate of the percent amorphous.
Preparation of known w/w percent mixtures of amorphous and crystalline material can also be used to check the PXRD method with the same caveats as presented for analysis using vibrational spectroscopy. For example, an amorphous standard prepared by ball-milling, spray drying, or lyophilization may not be fully equivalent to actual amorphous content produced in the sample by some other process (e.g. micronization using an air-jet mill or fast precipitation of the drug substance during crystallization).
Approach 2 – PXRD Quantitation of Amorphous with External Standards
As mentioned before, mixtures of amorphous and crystalline materials may be prepared. If a pure amorphous standard is not possible, a partially amorphous standard (e.g. 60%) may be used but would require estimation of the percent amorphous content by PXRD using Approach 1 or an orthogonal technique such as SSNMR. The same approach can be used if a pure crystalline standard is unavailable.
Each mixture can be prepared and then analyzed by method in Approach 1. The calibration curve would plot the percent amorphous, as determined by Approach 1 on the y-axis versus the weight/weight percent from the known weights of amorphous and crystalline material used to prepare the calibration mixtures. The calibration curve should be approximately linear allowing for a least squares fit. From the calibration curve, the observed % amorphous based on the area responses of the crystalline and amorphous components can be converted into a weight/weight percentage.
A simplified variation on Approach 2 is to use the peak intensity of a single XRD peak for the crystalline component and the intensity at a two-theta position where signal from amorphous content is present but there is not an overlapping crystalline XRD peak. The single XRD peak should be at a low enough two-theta value where significant signal from the amorphous is absent or the baseline intensity near the XRD peak can be used to subtract the amorphous signal from the XRD (crystalline) peak intensity. Preferred orientation and size of the diffraction domains of the crystalline component need to controlled [3]. With XRD deconvolution tools available, use of the full diffraction pattern is preferred.
With all the caveats using mixtures of external standards, the approach without standards is recommended. Many analytical results (e.g. % area-under-the curve by HPLC, particle size distribution) produce results that are relative and not absolute but still allow for batch-tobatch comparison and data trending.
Approach 3 – Use of Rietveld Analysis for Quantitation of Amorphous
If the single crystal X-ray diffraction structure is available, Rietveldbased methods that use the whole diffraction pattern may be employed to estimate the percentage of amorphous content in drug substance [2]. The approach would require spiking the sample with an internal standard where the crystal structure is known such as silicon powder. The data should be collected using a fixed divergent slit. The analysis involves calculating the diffraction patterns for the crystalline drug substance and the internal standard and then varying the relative amounts of each component until good agreement between the observed and calculated diffraction patterns for the spiked sample are obtained. The use of the internal standard allows the Rietveld analysis to provide the weight fraction of the crystalline component. Subtraction of this value from 1 would yield the estimated weight fraction of the amorphous component. Approach 3 can also be an alternate approach to check the results from Approach 1 or 2. The use of Rietveld approach for quantitation of amorphous should not be undertaken by the novice XRD user and thus no further details are provided. For the more sophisticated user, Rietveld tools are typically part of the software of modern research grade diffraction systems.
For samples with amorphous content below 10%, the difficulty in obtaining reliable results by PXRD increases and alternate techniques to assess amorphous content, such as gravimetric vapor sorption or thermal analysis, are suggested. For samples with crystalline material below 10%, Approaches 1 or 2 may be sufficient. For higher precision, standard addition methods by spiking the sample with crystalline material may be required.
Conclusions
PXRD represents a convenient method to determine amorphous content over a broad range. The use of the approach without standards is convenient and often sufficiently accurate to help drive the optimization of the drug substance synthesis.
Acknowledgement
The author thanks Graham Whitesell of GSK for reviewing this manuscript.
References
- Randall, C., Rocco, W., Ricou, P., XRD in Pharmaceutical Analysis: A Versatile Tool for Problem-Solving
- Powder Diffraction, Theory and Practice, Dinnebier, Billinge, S. (eds.), Chapter 11 (Quantitative Phase Analysis, authors Madsen, I, Scarlett, N.), The Royal Society of Chemistry, RSC Publishing (2008)
- A Practical Guide for the Preparation of Specimens for X-ray Fluorescence and X-ray Diffraction Analysis, Buhrke, V., Jenkins, R., Smith D. (eds), Wiley, VCH, 1998.
Biography
Peter Varlashkin received his PhD in Analytical Chemistry from the University of Tennessee (Knoxville). He has been employed by GlaxoSmithKline (GSK) for over 22 years. He currently works in the Physical Properties group within GSK (RTP, NC, USA). He has published several papers on powder X-ray diffraction and is on the organizing committee of the Pharmaceutical Powder X-ray Diffraction Symposia as well as a member of the International Centre for Diffraction Data.
This article was printed in the January/February 2011 issue of American Pharmaceutical Review - Volume 14, Issue 1. Copyright rests with the publisher. For more information about American Pharmaceutical Review and to read similar articles, visit www.americanpharmaceuticalreview.com and subscribe for free.