2D hybrid pixel detectors such as the Pilatus or Maxipix are essentially arrays of 0D photon counters. With these you get the best of both (or all 3) worlds.

I agree Carsten! Would there ever be any circumstances where a 0D detector could be preferrable to a 2D detector today? For instance, w.r.t. reciprocal space or real space resolution? Please post specs for the "Pilatus or Maxipix" when convenient. What is the typical cost point and the 2D image frame download rate?

Please post some of your relevant publications and 3D reciprocal space data when convenient. Could we assume that all 0D detectors at ESRF have been replace by 2D detectors? Thanks Carsten!

Please feel welcome to join & share your thoughts in this LinkedIn group discussion when convenient:

https://www.linkedin.com/groupItem?view=&gid=2683600&type=member&item=5893328475145662464&commentID=5894038264679014400&goback=%2Egde_2683600_member_5893328475145662464%2Egmr_2683600&report%2Esuccess=8ULbKyXO6NDvmoK7o030UNOYGZKrvdhBhypZ_w8EpQrrQI-BBjkmxwkEOwBjLE28YyDIxcyEO7_TA_giuRN#commentID_5894038264679014400

Pilatus! An awesome tool for XRD. What would that cost and what diffractometer would that transparently (through software) interface (adapt) to among existing XRD instrumentation?

http://www.slac.stanford.edu/econf/C0604032/talks/snic_broennimann.pdf

Medipix! Nice device. How much XRD software and support is available?

https://www.google.com/search?q=medipix&tbm=isch&tbo=u&source=univ&sa=X&ei=EHPCU42JAdezyASY64GgAw&sqi=2&ved=0CDwQsAQ&biw=1920&bih=851

In most cases you gain by using a 2D detector. However, more work is needed to analyse the data.

0D detectors are sometimes still used.

For example, for very low background measurements detectors with good energy resolution are used, e.g. silicon drift diodes.

For tracking very fast signals avalanche photo diodes are used. E.g. for pump-probe techniques such as nuclear resonant scattering.

I do not know how well these 2D detectors are integrated into the software of different suppliers. Our applications here are all very specialized and we use custom software.

These days the interface between detectors and diffractometer software is handled through an API called LIMA.

LIMA drivers for both Maxipix and Pilatus are available, as well as several CCD and CMOS detectors.

http://www.esrf.eu/UsersAndScience/Publications/Highlights/2011/et/et4

I will try to find some examples for you.

The Pilatus 6M is an amazing detector, but not cheap. Of course, the prices of these detectors are continually dropping as new models come on the market. In principle, any of these detectors can be interfaced to diffractometer software if you (or someone in computing) knows how to write a device server (like Taco/Tango). SPEC or EPICS are ideal for this, and the device servers for most detectors and motor controllers have already been written. Of course a license for SPEC is expensive ($5000 for academic institutes) but it is extremely flexible and used extensively, so there is a lot of support and software available freely within the community. CMOS detectors are also very good for low signal applications, since the noise is much lower than a CCD. They are also much more affordable than a Pilatus, and are ideal for high resolution camera systems. I've used a Hamamatsu CMOS to achieve an effective pixel size of .7 microns and a point spread of 2 microns.

0.7um is cool! 2.0um point spread. Does this mean one can identify a "Bright" pixel within a 2.0um precision? So, within 3 pixel or (3x3 pixel in 2D) precision?

What is the size and cost for this advantage? 12Bit, !6Bit or more?

BTW Larry, what is the smallest beam size you've had access to thus far for XRD observations? What is the typical synchrotron beam size in "point source applications?

Pilatus 6M - ID29 is now equipped with a Pilatus 6M-F detector which has an active area of 424 x 435 mm2 (2463 x 2527 pixels, which are 172 microns in size).

What is the fastest frame download rate for this beauty? How many Bits per pixel?

http://www.esrf.eu/UsersAndScience/Experiments/MX/About_our_beamlines/ID29/pilatus-6m-detector

Ravi,

Well, the effective pixel size is .7 um. if you have a reflection which is smaller than .7 um at the detector position then it will be recorded by the CMOS detector as a Gaussian with a 2 micron FWHM. This means that you can determine the COM position of the reflection to better than .2 microns (as long as you can properly correct for the spatial distortion with a spline to this resolution, which is limited to the precision of fabricating a grid mask). The Hamamatsu is 16 bit (digitized, true dynamic range is probably about 14.5 of so (it varies with the exposure time)). It's about 2560x2100 pixels. Unbinned max frame rate is 20 Hz, but you can pay more for a faster readout. It is an inline readout, so there is no deadtime. You can acquire the next image while you readout the last image without any readout streaking.

The smallest beam I have worked with was 25x25 nm at the nanoprobe beamline at APS. Focusing was done using bent laue multilayer lenses which were produced at NSLS-II for the nanoprobe beamline project I was working on during the design stages. That was at about 12 keV I believe. This is not easy to achieve. As far as typical synchrotron beam size, well I would say 1 micron to a few mm. 100-200 nm can be achieved very reliably on dedicated beamlines.

Detectors with very small effective pixel size are typically indirect detectors, i.e. a scintillator screen converts x-rays into visible light, which is then projected onto a larger pixel CCD or CMOS detector by a (visible light) microscope.

Carsten,

Yes, I forgot to mention that. We used a Gadolinium Gallium Garnet (GGG) phosphor doped with Tb to convert x-rays into visible light (550nm). The layer thickness is about 15 microns to minimize the point spread function. It was manufactured at the ESRF. To get a small effective pixel size we use a 5x or 10x objective lens to image the phosphor. A mirror to reflect the visible light into the optical lens to prevent the direct beam from hitting the detector. Ultimately the limit in spatial resolution in these detectors is due to the wavelength of the fluorescing light. Unfortunately, no one has yet come up with a direct 2D x-ray detector that is capable of sub-micron resolution.

Lawrence! "Ultimately the limit in spatial resolution in these detectors is due to the wavelength of the fluorescing light." I agree in terms of real space voxel size. Switching to P45 or P47 phosphor will bring it to the 400nm range. In reciprocal space on the other hand, the resolution is essentially unlimited since it is feasible to "rock" the crystal infinitesimally (delta Omega), Isn't it?

BTW would it also be possible using divergent beam optics and asymmetric reflection methods to probe the "sub-micron" range with lower resolution detectors. Besides, fiber-optic "reducers" and "expanders" may also be employed, couldn't they?

Hamamatsu and others are trail blazing in the high resolution image sensor technology. It is just a matter of time!

https://www.flickr.com/photos/85210325@N04/8473561270/in/set-72157645018820696

Ravi,

Yes Hamamatsu has the best low noise CMOS CCD high resolution detectors I've found. The low noise is extremely helpful in increasing the signal/noise and true dynamic range (rather than the digitized dynamic range, which isn't really a meaningful number explaining why vendors only quote this in their specs and force you to calculate the true dynamic range taking into account the dark noise, readout noise, and the linear full well capacity) since your phosphor must be extremely thin (5-15 microns) to achieve a small enough point spread which leads to low efficiency. Of course by using magnifying optics the effective pixel size of your CCD can get quite low so that the point spread function of the CCD is not limiting your resolution. It is ultimately both the point spread of the phosphor and the abberations of the optical lenses which limit the resolution in these applications. A divergent beam approach wouldn't work for 3DXRD since we rely on the shape of the diffraction spot being a projection of the grain shape. With a highly divergent beam the spot will be blurred and result in lower resolution reconstructed grain maps. In the end what we are interested in with what spatial resolution we can map polycrystals with 3DXRD. With the best detectors and phosphors today we hit a limit of 1 micron. I'm not familiar with P45 or P47 phosphors. We use GGG doped with Tb or Eu which are made at the ESRF with thicknesses down to about 5 microns I think. They are fairly new and have the best properties for use in the energy range we need (35-80 keV) in terms of light yield and especially in minimizing tails in the point spread function which can be a real killer for the reconstructed spatial resolution.

You are of course correct that I am talking only about spatial resolution. Reciprocal space resolution is limited by the resolution and wobble of your rotation stage but also in the 2theta and azimuthal position by the small sample detector distance (5-15 mm). This is why we construct the high res detector to be semi transparent (the thin phosphor and high energy then become an advantage since most of the diffracted intensity is transmitted through the phosphor and mirror. This allows us to have a large area detector (GE Si flat plate) far enough back to get excellent reciprocal space resolution. By coupling the data from these 2 detectors the spatial and angular information can be decoupled.

There are of course super resolution methods, but the software has not been developed fully yet. When that problem is solved it would be possible, in theory, to have a spatial resolution limited only by motion accuracy. The biggest bottleneck there would be the wobble of the rotation stage. I've seen air bearing rotation stages that claim a sphere of confusion of around 100-200 nm. In translation you can easily achieve 100nm and with a lot more work and money (a lot more money) down to 10nm. However I think at this point your spatial resolution would then be limited by the spatial distortion of the detector. It is not easy to get a spatial distortion spline with that resolution. I would even dare say it is impossible at these energies. I know at least one group has been working on developing a novel detectors based on CMOS technology but it would be a direct single photon area detector that could achieve 100 nm spatial resolution if someone could figure out how to fabricate one that actually is single photon counting with the flux from a synchrotron at these energies.