Measuring the anomalous differences in the Friedel pairs (or Bijvoet pairs) is the standard way to determine absolute structure (and I directly absolute configuration). You shoul collect data at low temperature and make sure you get sufficient Multilicity of Observations (MoO; sometimes called "redundancy") because the anomalous differences are usually weak and good and carefully collected data is required.

For light atom structures I recommend copper radiation as the anomalous signal from C, N, O and F is almost (but not quite) negligible with Mo radiation.

You may want to read some papers by Howard Flack, Simon Parsons and David Watkins in this context.

I was writing this answer in a hurry on my iPod touch. Here are some corrections:

"I directly" --> indirectly

shoul --> should

Multilicity --> Multiplicity

Do I understand your statements properly? Do you mean to say that the difference between F(hkl) and F(-h,-k,-l) will only show up if we are close to the absorption edge? If I am far away from absorption edges will they be exactly equal? I thought the single domain crystal should be enough to detect this difference in principle. Does the normal diffractometer experiment give rise to the inversion symmetry in the diffraction data?

No, you do not understand me correctly. Frankly, I am not sure WHAT you understood from my answer, but I recommended copper radiation to increase anomalous signal from light-atom crystals. With heavier atoms (such as S, P or Si) molybdenum radiation suffices and with really heavy atoms, like Br or so, the anomalous signal is so strong that it becomes difficult not to detect it.

With natural compounds that consist of C,H,N and O, the Bijvoet differences are quite subtle (maybe 1 or 2% of the total signal with copper radiation and much below that for molybdenum). That means you need to have a really low standard uncertainty to quantitatively detect anomalous signal. This is possible with low temperature data that has sufficiently high MoO and I do regularly in my own facility.

To summarize: there will be no difficulty to detect the anomalous signal if your crystal contains at least oxygen and you use copper radiation and collect data at low temperature (say 100K). If your crystal has only carbon and hydrogen, you still have a chance, but it is going to be difficult. If you have only molybdenum radiation at your disposal, it will be difficult to detect anomalous signal from oxygen and almost impossible for CH-only crystals.

Well I can do the experiment at the ESRF where I can go to the absorption edge of the elements that constitute the sample which happens to be magnetite Fe3O4 and also YCrO3.

I did not know you had a synchrotron at your disposal. I would not go to the absorption edge of any of the metals simply to enhance the anomalous signal. The anomalous signal of your crystal should be plenty strong at any suitable wavelength and if you go for a shorter wavelength you get more reflections (resolution limit is half the wavelength) with which you can do other things (like refine the structure against). Depending on the beamline (some are optimized for specific wavelength ranges), I'd go somewhere around 0.7 to 0.8 Angstroms.

In any case you'll get plenty of anomalous signal and, to re-address your original question, the diffraction experiment will not destroy the Bijvoet differences. Just collect a normal high-redundancy (>8), low-temperature (100K) dataset and you'll have all the anomalous signal you could want.

Thanks for your very useful advice. Although I am the neutron scattering center Institut Laue-Langevin (ILL) our next door neighbor is the European Synchrotron Radiation Facility (ESRF). So we can fortunately use both neutron or synchrotron for our research.

You have answered my original question. I was wondering whether diffraction experiment somehow introduce centrosymmetry in the measured intensities of a non-cetrosymmetric structures.

By the way I have seen your very helpful article "Practical suggestions for better crystal structures" in Crystallographic Reviews.

You can get this information through IPDS analysis. Either by transmission using a tiny crystal or backscattering mode for bulk sample.

Can you please explain this technique in more details. I do not know this technique neither do I know the full expression for IPDS.

Of course. This method is called Image Plate Diffraction System which the beam reach the sample (rotated) and is diffracted generating spots in the image plate/similar to the Laue Camara but for sucessive angles so that the full range is analysed.. and is relativelly simple. For each angle of rotation of our sample (phi) , a image is collected so that after a full range (complete rotation) the set of images, i.e. the reciprocal space image, may be analyzed by the program providing us the structure determination as well as the index of each image/each spot. If there is any assymetrie for any spot/hkl, after 180° (or not) it will be detected and analyzed by the software in terms of reciprocal space. Thus, you can obtain informations analyzing the images around the assymmetric center using very shor steps of phi. The software also give us information about certain assymmetries or deffects, quantifying the mosaicity for instance. You also can control the sample position for the psi angle - angle in a perpendicular plane whith respect to the phi angle plane. Currently I'm using this method to analyze a specific kind of diffuse scattering and extra-reflections for my sample. The description was focused in the transmission mode, for this the sample must to be thin (~100 micrometers) and considering the absorption parameters as well as used cathods as Mo, Ag, etc.

Thanks for describing the IPDS technique. Surely by this technique you can scan and measure intensities of the reflections very quickly and also measure Bijvoet pairs and check non-centrosymmetry. Thank you very much for the explanation.

Way above my pay grade! But I will learn about this. Thanks!

Check it out! This may be of interest to you folks:

"The 3rd Dimension and beyond in XRD 2013" by Weiguang Guan! How to use FREEWARE! August 21st 3D visualization of XRD - McMaster University. http://www.chemistry.mcmaster.ca/facilities/xray/221-max3d

http://www.linkedin.com/groupItem?view=&gid=2683600&type=member&item=264379262&qid=4c57cca1-128e-43ac-852f-d429b3025a71&trk=groups_most_popular-0-b-ttl&goback=%2Egmp_2683600

"I was wondering whether diffraction experiment somehow introduce centrosymmetry in the measured intensities of a non-cetrosymmetric structures. "

If the atomic scattering factor would be real (no phase shift), then there would be no difference between I_hkl and I_-h-k-l. But this is never the case. Around the absorption peaks the jump in the imaginary part is large (large contrast). For example the latest version of Vesta (http://jp-minerals.org/vesta/en/) can compute structure factors based on f' and f'' data from http://www.nist.gov/pml/data/ffast/. I would also recommend the program suite Demeter (http://bruceravel.github.io/demeter/) and the page http://skuld.bmsc.washington.edu/scatter/AS_Friedel.html (or the whole tutorial at http://skuld.bmsc.washington.edu/scatter/).

If you perform low temperature measurement you have to think about possible phase transitions, like the Verwey transition in magnetite.