The X-ray Fluorescence Microprobe at the Australian Synchrotron

XFM has two probes, differing mainly in their achievable spatial resolution. In brief:

• milliprobe: spot size range from 50-200 µm
• KB microprobe: spot size range from around 1 µm to 10 µm

In order to determine which probe would be best suited to your investigation, you will need to know the size of the smallest feature that you need to resolve.

We also have two fluorescence detectors. The main differences between these are (1) speed and (2) energy range (which determines which elements you can see). In brief:

• Vortex detector
  Dwell time will usually be around 0.2-2 sec per pixel.
  Sensitive to photons with energy above around 1.6 keV
  K-, L-, or M- lines of all elements heavier than Si (Z = 14)
• Maia detector (Rev C)
  Dwell times of 0.5 msec - 50 msec have been demonstrated
  Sensitive to photons with energy above 2.0 keV
  - K-lines of elements heavier than S (Z = 16)
  - L-lines of elements heavier than Zr (Z = 40)

While the Maia detector is extremely reliable, we note that it is a research detector and so it is not possible to carry a spare.  It is available on a best-effort basis, and every experiment should anticipate using the Vortex detector as a back up if needed.

If you are using the KB microprobe you will need to decide which fluorescence detector would be best suited to your investigation. This decision is a trade-off between elemental sensitivity and scan speed (which translates directly to increased measurement areas). Therefore you will need to know which elements you would like to map. If you do not need to see elements with fluorescence energies between 1.6 keV and 2.0 keV (i.e., Si & P), and are using the KB microprobe, then we recommend the Maia detector.

Estimating the scan duration

In short, the scan duration is determined almost entirely by the product of (1) the total number of pixels in the scan and (2) the dwell at each pixel.  The scan area is usually dictated by balancing your science goals, the instrumentation, the required resolution, and practical time constraints.

An (excel based) scan-time estimator can be downloaded from: scan_timing_0.xls.

Example 1

Let's say that you have two cell types with three treatments, and wish to have n=3 for each cell (giving a total of 18 cells).

Your study aims to investigate correlations between P and Zn, so you will need to use the Vortex detector.

The cells are about 5 µm in diameter, and you wish to obtain high definition images of internal distribution.  The ZP nanoprobe will be used at a target resolution of 400 nm.  Although the ZP nanoprobe can achieve a resolution of ~150 nm, the sensitivity at this resolution is not suited to mapping elements at biological concentrations.

Although the cells are 5 µm, scan areas of ~15 µm will be required to keep them within the target region, due to slow drifts in the system.

Scan overheads for this system are 500 msec per pixel, 550 msec per line.  The scan time calculated using the spreadsheet (don't forget to include the pixel overheads):

18 [cells] x 15 [µm wide] x 15 [µm high] / 0.0004 [µm x-resolution] / 0.0004 [µm y-resolution] x 2.5 [(dwell + overhead) seconds per pixel] = 18 * 62.1 minutes = 18.6 hours. 

You can use the 'stack time estimate' in the top left corner (yellow) to estimate the duration of multiple scans.  An overhead can be used to estimate the additional time required for cell finding and pre-scanning.  The 'skip' field can be used to ignore scans in the stack.

Estimating 20 minutes per cell for window mounting and scan area selection gives 24.6 hrs.  An additional 1 hour / day for measurement of the standards (+ 2 hrs), 4 hours setup (day 1; + 4 hrs), 1 hour optimisation (day 2; + 1 hr) results in a base request of 31.6 hrs.  If you are unsure as to your metals content, and so may need to explore increased dwell (up to 4 sec / pix), then this may require 39.5 hrs of minimum access.  3 days is recommended for this experiment, for new users or new (unknown) specimens.

Note that the dwell has been adjusted in this example by setting the value in the velocity field to a multiple of the step size: a velocity of 0.0002 mm/s with a step-size of 0.0004 mm is equivalent to a 2000 ms dwell, as reflected in column 1.

Example 2

Consider again the above example, but with two requirements modified: (1) no need to map Phosphorus and (2) prefer to map many cells at lower resolution so as to obtain population statistics rather than detailed sub-cellular structure.

S is now the lightest element, so you can use the Maia detector (Rev C).  For this detector the sensitivity per second is generally better, and so a dwell time of although some background in Fe, Cu and Zn can adversely affect maps of these elements. 

Resolution of around 1 µm is sufficient, so the KB microprobe is the instrument of choice.  Lower resolution generally means higher sensitivity, and use of the KB microprobe in preference to the ZP nanoprobe will greatly boost measurement areas.

Scan overheads for this system are 0 msec per pixel; 550 msec per line.  Long dwell times will be required for this kind of specimen; likely a velocity of 0.128 mm/s.  Note that, due to the line overheads, it is sometimes possible to trade sampling interval for stage velocity. The scan time calculated using the spreadsheet (don't forget to include the pixel overheads):

18 [regions] x 500 [µm wide] x 500 [µm high] / 0.001 [µm x-resolution] / 0.001 [µm y-resolution] * 4.69 [msec per pixel] = 18 * 62 min = 18.6 hrs.

How much beamtime should you ask for? It is difficult to say, as other factors such as specimen mounting and instrument access will come into play. You will need to become more familiar with the instruments in order to estimate these overheads, which could easily range from 10 minutes to 2 hours per specimen.  For a first visit, with (eg) 36 hours of direct measurement time required, I would suggest a request for three days to enable time for measuring standards (1 hour / day) and for beamline setup and optimisation (4 hours day 1, 1 hour / day thereafter).

Substrate selection

Two factors that influence the choice of substrate are:

• The substrate should not produce signal in excess of that expected from the specimen itself.
  For low-signal specimens (usually biological, but anything that is thin) this means it should be free of the elements of interest and should not produce significant x-ray scatter (this means that it should be as thin as possible, and made of reasonably light elements). For bulk specimens these issues are not so important, as the specimen itself will produce scatter. It is for this reason that thin specimens are generally preferred.
• The substrate should be relatively small so that we can:
  (a) bring the fluorescence detector very close to the specimen (useful for optimising fluorescence collection), and;
  (b) pack many close together on the specimen stage (useful for reducing change-over time).

Accordingly, we recommend;

• Silicon Nitride windows (see Silson at http://www.silson.com/index.html). These are particularly good as they come with field of view up to 5-10-20 mm (at a price).

Please contact us if you need a few windows to play with; we may soon be able to point you to a local manufacturer.

• EM grids (please be aware that metal constituents in the EM support frame may affect background, especially if the specimen is close to the gridbar or support)
• Formvar films - but how to hold them?
• LUXFilm (TM) - not yet tested, but these look promising (see Luxel at http://luxel.com/tem_supports.html). We could contact them to arrange metal-free supports.

Please let us know of any other mounts and suppliers that you think are good and we will add them to this list (and feel free to give us a sample of your favourite substrate).