Patents II. The Use of Powder Diffraction in Patents

Course Material Index  Section Index  Previous Page  Next Page

The Use of Powder Diffraction in Patents

Introduction

By presenting powder diffraction details in a patent the inventor can assert not only the novelty of a substance, but also the its "non-obviousness". (Consider to what extent you could predict the powder diffraction pattern and melting point of even a known molecule without knowing, for instance, the structure.)

Why powder diffraction for Patents?

There are a number of relevant features to note about powder diffraction:-

• Powder diffraction is a unique fingerprint method. The position of the peaks is given by the unit cell dimensions and the space group, and the intensities of the diffraction peaks come from the arrangement of atoms within that unit cell. Thus substances with identical (within error) diffraction patterns must be themselves identical.
• As far as a crystalline solid is concerned powder diffraction can be non-destructive to the powder ('here you see the very sample I used in these tests') and powder diffraction can be used on low quantity samples (say 10mg or less).
• Powder diffraction is very sensitive to the exact crystalline form (see the next section on polymorphism). It can look at the powder, e.g. a tablet exactly as it is sold, and minimising sample preparation is generally viewed as a way of reducing possible errors.
• Powder diffraction is based on rigid scientific principles and is not empirical or inferential like some other methods.
• Powder diffraction data can generally be presented in raw form, which may enhance its credibility.

However, powder diffraction needs a reference pattern, for instance from the ICDD or from the patent - establishing a generally accepted standard can prove difficult, as can comparison between different data sets. Pharmaceutical companies have been rather slow to submit their patterns to the ICDD, which is strange because it costs nothing and would establish a an effective standard.

Some tips and hints for powder diffraction work for use in pharmaceutical patents are included in the conclusion to this section.

What is preferred orientation?
Preferred orientation is the tendency of powders composed of crystallites which are either needle-shaped (acicular) or plate-shaped (platy) to line up in a particular direction, rather than randomly, when packed together (see earlier page on this subject).

What is the answer to preferred orientation?
The bad news is there is no single simple answer, but this need not necessarily be a problem as far as a particular patent is concerned. Even so, it may provide opportunities for cross-examination of expert witnesses in a court, since it means that the relative intensities in a powder diffraction pattern may depend on the data collection method as well as on the exact composition of the sample.

What can be done to alleviate the problem?
Grinding the sample is often given as the textbook solution, but caution is advised in pharmaceutical patent cases because of the possibility of a phase transformation (or the somewhat larger possibility that robust cross-examination will suggest the possibility of phase transformation!).

Check the sample first with a microscope (or consider electron microscopy). As a first rule of thumb if you cannot see any strongly non-equant (platy or acicular) habit with a powerful light microscope then preferred orientation will probably not be a problem. You can also have a quick look to see if microscopy indicates that the sample looks to be of a pure single crystalline phase. Using a polarising microscope also means you may be able to check crystal optics at the same time.

Take care over the sample preparation: if you do suspect preferred orientation then packing from above and smoothing down with a glass slide may only aggravate that. Consider side filling or sprinkling onto a substrate (with a very thin layer of silicone grease for instance). A recent ad-hoc round-robin by the Industrial Group of the BCA showed that the latter may not work well in practice. Try to be consistent with your sample preparation because at least that should make it easier to make comparisons between your samples. Sample Preparation is covered fully in earlier parts of the course.

Always consider collecting a pattern before grinding any sample. If the data quality is acceptable all is well and you cannot be accused of having induced any changes. Then consider grinding the sample and re-running it. Changes in peak intensity imply changes in preferred orientation. An entirely different pattern may dindicate a phase change.

In a platy sample the intensities of reflections from the large faces may well be enhanced, so if you find (100), (200), (300) reflections all stronger than anticipated you well suspect preferred orientation.

If data can be collected in both reflection and transmission geometry, then do so. This is the best method for both checking and even measuring the preferred orientation.

A number of powder-pattern simulation and Rietveld and similar programs have parameters to correct for preferred orientation and these can be used to make comparisons, but this should be considered a method of last resort. Most are unsatisfactory in some way.

Indexing and Unit Cell Refinement

These have been fully covered previously in the course, and you may want to go back to check the Principles of Indexing section and/or the Unit-Cell Refinement section (but please be sure to return here to continue this section). These previous sections on this course are properly mathematical and comprehensive; however such detail and equations rarely impress lawyers or courts so I present below a "words-only" version, slanted towards use in patents. Indexing of powder diffraction data is extremely useful, even essential, when the question of sample purity is raised. When asked on the witness stand whether a powder diffraction pattern that has been collected many times with significant variation is from a pure single phase, a surer answer is possible when the pattern has been indexed. For example, one might be asked: could this consistent pattern be a mixture of two phases, which always appear in the same ratio as the products of a reaction? or could there be any minor impurity in this sample?

What is indexing?
Since the position of all the peaks in an x-ray powder diffraction pattern are derived from the unit cell, we should be able to go backwards and derive the unit cell from the position of peaks in a powder diffraction pattern. This is not always as easy as it sounds, and is very difficult with a mixture of two unknowns of low symmetry. Note that intensities are not important in indexing, beyond deciding whether a peak is detectable above the noise and and whether its position can be determined precisely enough. A single pure polymorph or phase only needs to be indexed once. The unit cell is clearly the same (within the usual minor variations, which can be accommodated by cell refinement) for all time.

What is required to index a powder diffraction pattern of a sample where the unit cell is unknown?

A Pure Sample (although this may be what you are testing for!).
If you can index every line and refine the unit cell you can usually be sure you have a single phase or polymorph (though this is less certain if the sample is a metal). If you cannot index a pattern, then you cannot say anything about its purity on the basis of your diffraction data alone. It is generally felt that if a good quality pattern cannot be indexed, this may well indicate that the sample is not a pure single phase. You can often cope with a small amount of impurity where the diffraction pattern of that impurity is known, simply by ignoring the peaks from the impurity.

A Diffraction Pattern
Running a sample with the sole intention of indexing can be a little different from standard data collection. You must have as small a 2θ zero error as possible, and know the precise value of it. Careful calibration, with for instance using the NIST low 2θ standard, is recommended. You need the highest resolution; and in the first run concentrate most of your counting time to cover the first 25 peaks or so (because most programs start by taking just the first 20 peaks).

Data Analysis Software.
Indexing will generally entail the following steps, for which an appropriate program or suite of programs will be required. First, the corrected 2θ positions of at least the first 20 peaks should be determined as accurately as possible. Standard peak location programs used to search/match are not always accurate enough, and fitting the peaks with a suitable peak shape function may be preferable to graphical estimation. An indexing program can then be used to derive a list of possible unit cells more-or-less consistent with the d spacings corresponding to the given peak positions. Most indexing programs give a figure of merit as described earlier in the course for each proposed unit cell. As a rule of thumb, a figure of merit below 4 means that the solution is useless. If it is around 8 the solution is certainly worth considering if you have indexed all lines, and if it is 16 or greater the solution is probably the correct. Possible answers are often ranked with the highest figure of merit first. Checking the crystal density can be most helpful at this stage. Taking the number of asymmetric units per unit cell as the number of molecules (or more correctly formula repeats) per unit cell, you can calculate the density and compare it with that which you might expect. For instance, a density of around 1.3 to 1.5 would be typical for a pharmaceutical substance, in which case a figure of about half this would indicate that there are two molecules per asymmetric unit; remember that some residual solvent of crystallization may be present in the sample.

Cell Refinement:
If you have a possible unit cell from powder diffraction data, from a single crystal structure determination or from an ICDD entry, you then merely need to refine the cell and confirm the purity of your sample. If necessary you should then make a further data collection covering an extended 2θ range, so as to include as large a set of d spacings as possible, and further refine the unit cell you previously found and check that all the peaks are in the expected positions. Systematic absences may then indicate the space group. Be sure to include the unit cell parameter values in your submission to the ICDD. This is rightly held to be a sign of good quality data. Don't forget that the pattern may show preferred orientation effects.

Course Material Index  Section Index  Previous Page  Next Page