Taking fluorescence tomography images Matthew Marcus

May, 2012

1. Intro

There is a setup for taking fluorescence tomography (f-tomo) images of small

samples, one slice at a time. This setup consists of a stage arm bearing an air-bearing

rotation stage, plus a small Huber goniometer head for the sample and another which

holds an alignment fixture.

The principle of f-tomo is that one takes linescans across the sample at a number

of rotation angles. The rotation axis must be perpendicular to both the scan direction (X

of the stage) and the beam direction. Thus, when installing the f-tomo stage, it is

necessary to adjust the tilt of the air-bearing stage. If the axis of rotation is off by an

angle δθ, then there could be an image blur of amount wδθ, where w is the width of the

scan.

In addition to acquisition of fluorescence data, it is useful to map the transmission

through the sample at the same time. This is done using the 8’th channel mechanism,

with the transmission ion chamber behind the sample.

II. Installation

The procedure for installing and aligning the tomo stage is as follows:

1. Install the stage. The tomo stage lives in the hutch, under the table, and has all its

cables already connected. Replace the normal stage arm with the tomo stage, as

shown in Figure 1. There is a level attached to the top of the arm, which you can use

to get the stage close to alignment. Don’t worry if in pushing and pulling on the stage

you break the servo lock on the X stage and it ‘goes limp’. It’s actually useful to

have it in that state for the alignment process.

2. Get the air-bearing working. On the upstream wall of the hutch is an air-handling

setup with an air inlet coming in from the left, a gauge on the right, and a line coming

off to the right. A metal pipe on the left has a red-anodized sleeve. This is the inlet

valve. Push this to the right to turn on the air. You should hear a short hiss. Now the

air bearing should spin freely. Do not turn the stage without air.

3. On the control computer, choose Motors->Set Up Parameters. On the screen

which appears, there is a list with a white background and a scrollbar. Scroll to the

bottom, and you will see Air Bearing Theta as one of the displayed motor names.

Choose this one. Then press the Motor Used? and Display in List? buttons. Next,

choose Function->Save from the menu bar in this screen. Dismiss this panel. The

Motor Display.vi panel will now expose Air Bearing Theta as one of the available

motors, in the All group. Verify that you can move this motor. Now, hit the Disable

button so that it spins freely again. If the controller (white box with a red EMO

button in the Bruker AXS half-rack) has been turned off, then the motor may not

work. To get it working, hit the Advanced>> button on the Motor Display panel.

There will now be a set of green lights showing at the bottom of this panel. The one

labeled Motor Off will be lit. If so, then choose Home Motor in the Motor Type

menu exposed when you went to Advanced mode. This should clear the logjam. Go

back to <<Standard mode.

4. Install the alignment head, shown in Figure 2. This consists of a Huber head topped

with a flat disk, near the edge of which a pin sticks up a little from the surface. The

first step in alignment is to make the surface of this disk perpendicular to the rotation

axis. To do this, set up a dial indicator as shown in Figure 3 so that it touches down

on the disk near the rim. Rotate the air-bearing stage and watch the indication. You

want the indication to remain steady to within a few microns. If it moves in a

consistent way on rotation, then adjust the tilt cradles on the Huber head to make that

effect go away. If it is possible to get the use of an autocollimator, this adjustment

may be performed off- line (stage not installed in hutch) using an accurately-flat

mirror, which we have. That style of adjustment is more accurate than the use of the

dial indicator.

5. The next step is the adjustment of the parallelism of the stage motion with the surface

of the disk, which implies that the stage motion will be made perpendicular to the

rotation axis, by virtue of the previous step. To do this, set the dial indicator so that it

touches at the center of the disk and move the stage horizontally by hand back and

forth, watching the dial indicator. To be able to move it by hand, you need to break

the servo lock, which you can do simply by pulling on it with enough force to make

the servo motor give up. You may have already done this accidentally during

installation. The reading will probably change. To adjust this, use a 5/64” Allen

wrench to tweak the adjustment screw called out in Figure 1 as #1. Try steps of ¼

turn. If you get it accurate to 5µm over a 2cm travel, then a 1mm object will be

imaged with a blur due to this error of less than 5µm x 1mm/2cm = 1/4µm, which is

good enough.

6. The above steps could be done without beam. The next requirement is that the

rotation axis be perpendicular to the beam. This is where the pin on the disk comes

in. The idea is to find the height at which the pin blocks the beam with the pin

upstream of the disk center and downstream, and make those two heights equal. To

perform this operation, it is first necessary to move the stage horizontally so that the

beam goes over the center of the disk. Place a small object on the center of the disk,

which can be seen via some machining marks. A 0-80 or 2-56 hex-head bolt, head

down, works well. Focus on this (roughly) and move the stage so that it’s in view,

just as you would if it were a sample. At this point, you will need to re-enable the

stage X motion. If you’re lucky, hitting Enable on the Manual Stage Control will

do it. If not, you may have to hit Enable, wait a few seconds, then hit Home, after

having moved the stage back so that it won’t hit anything on going through the Home

ritual.

7. In the Manual Stage control, set the Slew and Retrace speeds to 1000. The tomo

stage is heavy, and slow slews will induce less vibration than the default.

8. Now let’s say that you have the stage in the right place. Set up the transmission ion

chamber, with offset turned off (control computer) so that blocked beam = 0

transmission. Take your centering object off. If you see transmission, move up in Y

(Y-) so that the transmission goes away. Otherwise, move down in Y. You should

see the rim of the disk blurrily visible on the sample camera, so you can see if the

height is about right. Find the height at which the beam is just blocked by the disk

and then move down 100um. Beam should now be transmitted.

9. On the control computer, use the Single Motor Scan menu item in the Beamline

10.3.2 panel to run a scan of channel 1 (transmission) vs. Air Bearing Theta. The

parameters are: Motor to scan: Air Bearing Theta. Start = 0, Stop = 360, Step = 5deg,

Count Time = 0.5 sec. You should see two sharp dips, which represent the pin going

across the beam, upstream and downstream. These dips should be about 180deg

apart. If they aren’t then the horizontal position isn’t right and the beam isn’t going

over the center of the disk. Move the stage left or right using the set stage position

program so that the angular difference is 180deg. If you don’t see any dip, move the

stage up some. Write down the angles for the bottoms of the two dips, which for now

we can call θ1 and θ2. Figure 4 shows overhead views of the alignment head with

the angle set to θ1 and θ2 so the pin blocks the beam from going through to the

transmission ion chamber.

? 1 ? 2 horizontal movement delta ?

a 10 175 starting position 165

b 10 175 +100 165

c 15 175 +500 160

d 10 175 +500 165

e 10 180 -1600 170

f 5 180 -500 175

g 5 183 -500 178

h 4.5 182 -100 177.5

i 4.5 182 +50 177.5

j 5 182 +200 177

k 4.5 182 -150 177.5

a. moved in +x direction 100 microns. No change in theta.

b. moved in +x direction 500 microns, and to give faster scan changed the

degree steps from 2 degrees to 5 degrees. no change in theta.

c. moved in +x direction 500 microns, theta change less than 180. total move

was 1100 microns, need to move back in other direction towards centre of

axis of rotation.

d. moved in -x direction 1600 microns. Theta back to near starting position.

e. moved in -x direction 500 microns

f. moved in -x direction 500 microns. good, now move finer resolution angles

(change from 5 degree increments to 1 degree increments).

g. moved in -x direction 100 microns, this will give us a theta value that we can

relate to our horizontal micron adjustment.

h. moved in +x direction 50 microns, no change in theta

i. moved in +x direction 200 microns, oops worse.

j. we are in a current position of ‘+150’ relative to our best position which was

178 degrees (movement ‘g’).

k. moved -150. back to 177.5, good enough, enough time spent!

10. Use the Motor Display to move the Air Bearing Stage to one of those dip angles, say

θ1. Find the height at which the pin just cuts half the beam. You can do this

manually or using the Knife-Edge Scan program, just as if you were tuning the

beamline. The FWHM will be relatively big because the pin will be out of focus in at

least one of the positions. Do the same for θ2. If you come within a few microns of

the same value, you’re done.

a. move to 182 degrees

b. use knife edge scan program

c. scan in y axis

d. slits: h=300, vertical=20

e. FWHM: 15 x 3

f. on the MCA, ensure centre position given in ‘set stage position’ is loaded into

‘knife edge scan’ in the ‘scan’ box. we have it set for 25 microns x 1 micron

steps.

g. only got half a scan, set 50 micron x 1 micron steps.

h. ROI count at max peak were around 26000, so we want 13000.

i. ‘set stage position’ moving the stage in y direction, ‘jog axes’to acquire half

beam strength

j. FWHM value of result is10.7049 at 182degrees (theta2).

k. in motor display move to 4.5degrees (theta1)

l. at 4.5 counts were 24762 ROI

m. repeat scan, 33.2260 so we need to move (adjust screw below)to achieve

around 10.70 plus/minus 5 microns or so: 5.7 to 15.7

11. If you don’t get the same value for both angles, adjust the screw shown in Figure 2 as

#2. Try a 1/4-turn adjustment. Repeat Step 9 and iterate until the two height values

are the same to within 5 microns or so. The reason for using screw #1 for the

previous adjustment and screw 2 for this one is that #2 does not affect the parallelism

of stage motion and disk, while #1 affects both of the things you want to adjust.

FWHM ? 1=4.5 FWHM ? 2=182 degrees of rotation

moved

33.2260 10.7049 initial position

48.3611 11.4766 45 anticlockwise

21.357 11.0765 90 clockwise

15.820 12.7941 +45 clockwise

You’re now done with the tomo stage installation and alignment process. Time to

install a sample!

III. Sample mounting and alignment

Mount a sample on the sample goniometer head as shown in Figure 5. You can

use putty to attach it as shown. The region of interest should stick up to about the height

of the disk on the alignment device. If it’s too low, the gonio head will bump the

connector on the bottom of I0. If too high, the sample might be mechanically unstable.

Carefully put the alignment head back in its case and put it away. If you use the Cryo-

Stream to cool the sample (we haven’t tried this yet) you probably want it high. You

should be able to see the sample on the sample camera. If you don’t, move the stage so

you do. Now, Disable the Air Bearing Theta motor so that you can spin the rotary

stage by hand. The sample will appear to sweep back and forth as you do, and also wag

back and forth (position and orientation). To fix the wag, you can adjust the tilt cradles

on the gonio head (the lower ones) so that the axis of the sample stays vertical as you

rotate. This adjustment has a range of 30o in each direction, but if you use that range, the

cradles stick out far from the center, making it necessary to increase the height of the

sample over the head so that the cradles don’t bump the connector at the bottom of I0

during rotation. Some samples don’t have an obvious axis, in which case leaving the

cradles centered is good. After that, you need to adjust the centering, using the small

translation stages at the top of the gonio head. The aim is to make the area of the interest

of the sample sweep through as small an area as possible as you rotate so as to minimize

the required length of the scan lines in your sinograms. By spinning the stage, you can

see that the sample will appear to rotate around a center. Note where that is. Find the

angle at which the sample goes farthest to the left and adjust one of the translation screws

to move it partway to center. Find the right-most extreme and do likewise. Keep

iterating until the sample spins within the smallest area. This will take some iterating

until you get the feel of it.

Next, re-Enable the Air Bearing Theta motor and focus on the sample as you

would for ordinary XY mapping. The angle will probably read some large value because

you spun the stage around during alignment. Move it to 0 just to unwind it. Set up the

transmission chamber so that with the sample out of the beam you get 1-2 units of signal

with the sample out of the way, with the offset turned off. Make sure the short cable with

the pink label on it is plugged into the XRF hole so that transmission will be mapped.

Also make sure that the I0 offset is correctly set as neither the transmission nor

fluorescence channels will normalize out changes in the beam if that’s not done. The

offset for the transmission channel should be turned off. This is done on the same

computer you use to set slits. From the Beamline 10.3.2 window, choose Amplifiers-

>Set up parameters. This will bring up a window. In that window, click on Sample

in the Amplifiers list (upper left). Set Input Offset Current State to Off (top row of

controls). Choose Function->Save.

Set up an XY map which encloses the area of interest and includes the

transmission ROI as well as ROIs for all elements of interest. Don’t forget to choose an

appropriate scan energy. This map will help you decide on the slice(s) at which to take

the tomogram and also show whether the transmission got properly recorded.

IV. Tomography at last

We’re almost ready to take the sinogram. After choosing the slice, it’s necessary

to find the left and right limits beyond which the surface of the sample doesn’t go at any

rotation angle. The XY map you just took is at a single angle, so the X range is a subset

of the X range which will need to be scanned. With the XY map loaded, pick your slice,

put the cursor at that level, at the left edge of the sample, and hit Move to Cursor. Now

go to the Manual Stage. There should be a page visible called Theta. Click on that to

expose it. This gives you the ability to control the rotary stage from the data-taking

computer. Start the MCA. Move the sample to the right (X+) until you find the left-hand

edge, as seen from the MCA signal. The Theta page of the Manual Stage has jog

buttons labeled Th- and Th+. Set the Theta stepsize to 10. Jog theta until you see

signal appear on the MCA. Move the sample to the right until this signal goes away.

Jog theta some more. Keep jogging and moving until you can go a full 360 without

seeing significant signal. Move theta to 0 (you can do that on the control computer) and

hit Get Start Position. Next, move over to the right-hand edge of the sample, and perform

the same theta-X jog ritual, only this time moving the sample to the left (X-) when signal

is found. There may be some confounding signal from the “ghost beam”, but you should

be able to figure out where the real edge is. Once you can again go 360 without hitting

the sample with beam, move theta to 360 and hit Get Stop Position.

Now, set up a check sinogram. This is a sinogram with a coarse angular step,

intended to preview the real one, find problems, and verify that there is blank space

around the sample at all angles. Push the Scan Params button in the mapper and hit the

To XTheta menu item. Do the usual stuff with filenames and scan numbers, and edit the

scan. As in a normal map, hit the Load From Manual Stage button and select the pixel

size you want. On the right-hand side of the screen, there appears a set of controls for the

angle. Set the Start to 0, the Stop to 365, and the Step to 15 (we can use 30 theta). This

makes for a coarse sinogram, but fine enough to show you what you need to see. Set the

energy as appropriate on the Energies tab, and the ROIs, which should include one for the

transmission channe l. The values here are typically 1687-1874, but you can verify by

firing up the MCA utility and loading the 8 channels configuration. The transmission

signal will appear as a peak on the right-hand side of the spectrum. Exit the Single-

Scan editor.

On the main window of the mapping program, below the ROI display, there is a

small tabbed control. One of the tabs is labeled “Det”. To make sure the transmission

signal gets recorded. you need to set the Det. config. to 8 channels (vertical switch)

and the array of Special ROIs to have as its sole member the ROI you’re using for

transmission (usually 0). Labeling an ROI as “Special” tells the deadtime correction

program to treat it differently from ROIs derived from the detector.

Now, in the Multiscan Editor, hit Move To Start then Move To Stop, and you

should see the sample move right and perform a pirouette. Move to Start again and start

the map going. When this map is done, you’ll be able to tell whether you need to move

the Start or Stop limits out or if you can pull them in a bit, and how good the signal is.

Note that the reconstructed image will always look noisier than the sinogram. While it’s

normal to go around a full 360o as suggested above, it is in theory possible to get a good

reconstruction with only 180o of data. This may be necessary if something bumps when

you try to go all the way around, or if the sample geometry is such that going 180 will

result in a lot of empty space in the sinogram.

Once the check sinogram is done and you’re satisfied, you can set up the real

thing. You want the number of angles to be at least equal to the number of pixels across.

You may be shocked at how long it will take. You should be ready to start the sinogram

going and, at long last, acquire a tomographic image.

V. Data reduction

The data are in the same form as any other XRF map, and the same programs read

sinograms as other .xrf files. The first step in data reduction is deadtime correction,

which is done by the obvious program. If the Special ROI hasn’t been assigned, you can

fix the .xrf file by opening it in a text editor and adding a line

“Special ROIs: 0”

just under the one that starts “Log(I0 gain)”.. If the transmission ROI wasn’t the

first (#0), substitute its number for 0 in the above.

Next, run the XY Display program. Check the transmission channel. It will

probably be full of “speckles”, which are white-black pairs of pixels scattered randomly

over the image. These are the result of some flakiness in the detector electronics, and

may mostly be corrected using the Despeckle button in XY Display. Set the Upper

threshold to 1.2 and the Lower to 0.8, as a first try. Use the Save Masked Image button

to save the result.

If you want to see the optical density, you can use the Log scale function to

convert the transmission. First, use Lineout to find the average pixel value in an area of

the image which is outside the sample. You can Copy that value from the indicator in

Lineout and Return from the Lineout window. Next, enter the pixel average into the

Baseline control on the Log scale tab and set the Scale factor to 1000. Make sure

you’re looking at the transmission channel when you do this. Now hit the -ln(current

SCA) button. The image should change to one in which the pixel values in the blank

regions are near 0. You can read off the optical density by reading the pixel values (with

the crosshair) and dividing by the Scale factor. Thus, if the pixel value is 200, the OD at

that point is 0.2 for the incident energy. The reconstruction programs do their own

version of this procedure, so you don’t need to save this version of the file.

X motion

Stage arm

Rotary stage

Goniometer head

2 1

Adjustment screws

Rotary stage

Level

X motion

Stage arm

Rotary stage

Goniometer head

2 1

Adjustment screws

Rotary stage

Level

Figure 1. The tomography stage mounted, with the sample goniometer head. A sample is mounted on the head. The transmission chamber would normally be set up downstream

of this setup.

Disk

Pin

Tilt cradle knobs

Disk

Pin

Tilt cradle knobs

Figure 2. The alignment head. The disk is made perpendicular to the axis of rotation

using the tilt cradle knobs, while the adjustment screws on the stage arm (Figure 1) are used to make the disk parallel to the X motion and the beam.

Figure 3. The dial indicator set to measure the parallelism of the disk with the X motion. For the previous step, in which the disk is made normal to the axis of rotation, the finger

of the indicator would be set closer to the edge of the disk.

BeamBeam

Figure 4. The alignment head in the two orientations in which the pin blocks the beam. The height at which the pin blocks half the beam should be the same at both angles;

that’s when the stage axis of rotation is perpendicular to the beam.

Centeringknobs

Tiltknobs

Sample

Centeringknobs

Tiltknobs

Sample

Figure 5. A sample mounted on the sample head, showing the adjustments for tilt and

centering. The actual sample is the black object; it’s glued to a toothpick for mounting.

Sample mounting, analysis and data processing

Dave McNear & Alison White. October 2013 Mount a sample on the goniometer

1. Insert capillary into putty on centre of goniometer head. 2. Visually align parallel in the vertical. 3. Release the air-bearing on goniometer by pushing to rotate, and spin to ensure

parallel over 360 degrees.

Center sample on goniometer axis 1. Spin the sample and mark with a highlighter on the camera screen in the hutch

the outermost positions of capillary translation. 2. Using the small socket adjust the nut that is on the side of the goniometer and

perpendicular to camera view so that the capillary moves equidistant between the two marks on the screen.

3. Iterate with decreasing range of translation until no translation occurs when rotated.

4. The sample is now in the rotation centre of the goniometer.

Re-enable motor (air-bearing theta) 1. The air-bearing rotary stage releases when mounting and centering new

samples after which it needs to be ‘enabled’ before proceeding with mapping or tomography.

2. One way of knowing if the stage is enabled is by looking at the SetStagePosition panel and looking for the “Theta” tab in the upper left quadrant of the screen (to the right of the XY tab).

3. If Theta is not there then go to the control computer by hitting ‘scroll lock’, ‘scroll lock’ and then ‘enter’ (you will hear beeps)

4. Find the motor display panel and select ‘All’ from the Group to View pull down menu. Then select Air Bearing Theta from the Motor drop down window.

5. Now depress the ‘enable’ button to enable the air bearing stage. Enter 360 degrees in the upper white box and hit move to bring the stage back into range.

6. Verify that it is going back to position by looking in the top right video monitor and look for the stage spinning.

7. Moving the stage back to some value between 0 and 360 at this point stops it from winding back to range when you go to start an actual 2D map for positioning which can screw up data collection.

find edges of capillary in camera view

1. Use the MCA ‘multipoint MCA utility’ to find the edges of the capillary.

2. To turn on (it was ‘suspended’ because we were previously mapping) click the “switch” beside the big red STOP button and it will flip up and show ‘run’. You will know it is running because the peaks will start jumping around and the green button in the upper right quadrant will blink ‘counting’,

3. Watch for ‘scattering’ peak from the edge of the capillary (around 1000) in the spectrum window. What is observed from the camera is not in alignment with the peaks, so move in X until counts are observed, this is the edge of the capillary.

4. Go into hutch and move sample so that the capillary edge is in vertical alignment with cross hairs on screen.

2D map – set start and stop positions using SetStagePosition.vi GUI

1. Maps are collected starting at the upper left hand corner and finishing in teh lower right hand corner. To set up a map, these positions must be selected manually using the SetStagePosition GUI.

2. first start by moving to the upper left hand corner of the object you wish to be mapped. Move slowly to the right (-X) watching the multipoing MCA utility until you see a peak appear around 1000. This is the edge of the capillary.

3. Establish where the edge is using the MCA utility and then go ~100 microns away from the edge. Press the “Get Start Position” button on the SetStagePosition window.

4. Now move to the lower right hand corner of the object you wish to map and do the same thing as before, except this time find the edge of the capillary and move to the right (-X) about 100 microns. Press the “Get Stop Position” button on the SetStagePosition window.

5. Now test that the start and stop position are in upper left and lower right, respectively, but pressing the ‘move to start’ and ‘move to stop’ buttons on the SetStagePosition panel and watching in the right video monitor (or left if you have the detector pulled out and can see the sample) to confirm that the start position is in the upper left and the stop in the lower right.

6. Now you are ready to load these positions into the mapping program 2D µXRF map – set and load map parameters

1. Go to the XY or XE multiscan_combine loops panel which controls the acquisition of 2D maps and tomograms. go to mapping program

2. Select green “Scan Parameters” button on the left side of the screen. A window will pop up into which you will type the name of the map being collected

3. Select the ToXY option from the menu bar and another window opens into which the various scan parameters are entered.

4. You will see a box blinking red prompting you to load from manual which is referring to the start and stop positions you manually entered previously in the SetStagePosition panel. Push this button and confirm that the start and stop positions are the same as those on the SetStagePosition pane l.

5. Enter the desired scan parameters (usually just changing the X pixel size) to get the desired image resolution/scan time.

3000 x 1100 um map. slits: 100 x 20 ;6 x 3 FWHM (horizontal x vertical) pixel size x=10, y=20. this takes around 8 mins. select ‘scan parameters’and the box opens. edit scan , opens stage parameters, set all, everything should ‘load from manual’, just check. return. second tab, enter titles for files (sample identification and‘2D map test’). to overwrite previous scan take ‘scan number’ back one; otherwise the program automatically writes to next number. return. run. if not enough image beyond edge of capillary, then abort, and increase start and stop values in x the required amount.

2D µXRF tomography – set and load parameters 1. On the ‘XY or XE map, multiscan_combine loops’ dialog box, right click the

small green cross under the map panel and select ‘bring to centre’ which provides a x and y axis for setting the position and ends of the beam transect (slice)of the sample.

2. Select the position of the X axis by examining the distribution of each element of interest.

3. Now position the Y axis about 75 – 100 microns from left side of the capillary edge. Select green button ‘go to cursor position’. You should see the sample move in the camera. this is our start position.

4. in the ‘set stage position’ dialog box, select ‘get start position’. 5. Now position the Y axis to where we want the stop position; about 75 – 100

microns from right side of the capillary. Again, ’go to cursor position’. Now ‘set stop position’ in the dialog box ‘set stage position’.

6. Check that Y values (degrees) are the same, and X values are what we want. 7. On the ‘XY or XE map, multiscan_combine loops’ dialog box, select ‘scan

parameters’. 8. In the ‘Multiscan editor simplified’ dialog box, select the ‘XY’ tab then press

square button ‘load from manual’. Return. 9. Now to set up the tomo stuff. 10. Select the ‘Xtheta’ tab, then press square button ‘load from manual’. 2D µXRF tomography – test sinogram: 1. Now we set up the parameters for a test sinogram. This is to ensure the scan

covers 75–100 microns wider than either side of the capillary; this boundary region is essential for tomograph reconstruction.

2. pixel size: 10 x 20 (5 x 20 is ok for higher resolution test). 3. theta step: 15 or 30, for 360 theta. 4. theta start scan: 0; theta stop scan: 360. 5. 50 ms dwell, 500 ms settling

6. click on light grey region outside parameters box to update ‘line scan time’ details. 3 to 10 mins usually, depending on sample size. Return.

7. In the ‘Multiscan editor simplified’ dialog box, select ‘file’ tab. Change file name to have ‘_sinogram’ eg. ‘3b_WT_highZn_topnodule13_sinogram’. Ensure the number in the green ‘scan number’ box is zero. Return.

8. In the ‘XY or XE map, multiscan_combine loops’ dialog box, click ‘Run’. 9. If insufficient region appears outside capillary, select ‘Abort’. 10. Go back and increase X values the required number of microns.

Load into Xtheta, check the time it will take. Return. Rename file (set scan number back to zero). Return. Run.

2D µXRF tomography – run tomography 1. In the ‘Multiscan editor simplified’ dialog box, select ‘scan parameters’. 2. rename file to end with’_slice’ eg. ‘3b_WT_highZn_topnodule13_slice’. 3. check scan number is zero. 4. return. 5. select tab ‘edit lines’ 6. change parameters: 7. pixel size: 3 or 4 um X; 20 um Y 8. dwell time 50 ms; settling time 500 ms. 9. theta start scan: zero 10. theta stop scan: 360 11. adjust theta step to 0.5 12. click outside light grey box to check how long it will take. 13. return. 14. return (red button). 15. Now, ready?, cross fingers and click ‘Run’. Go for food!

Tomogram Reconstruction 1. Create new folder for each slice file (sinogram) 2. Copy the slice file into the folder along with the Fiji routines a. i.exe b. op.exe b. wt.exe c. xrf.java d. xrf.class 3. Open Fiji, drag and drop the xrf.java routine to start the program and the program window will pop up (Figure 1) 4. hit the run button in the lower left quadrant

Figure 1

5. you will be prompted to open the sinogram file. Find the directory where the files are saved and chose appropriately. 6. Once a file is chosen, the XRF conversion window pops up (Figure 2). The first box is requesting which row of data contains the transmission file. The second box is requesting the location of any data that you don’t want processed. The first check box is to normalize the data by I0 which should be done. The second is to smooth I0 before using it to normalize the sinogram. The second box is there to remove outliers (e.g from Bragg diffraction) and allows thresholds to be chosen. Go with the defaults. Hit OK 7. Once the file is converted a new window pops up (figure 3) which will be used to align each of the slices for reconstruction. The best way to start is to select a range from 0 (low) to 12 (high) (or 0 to -12) with a step of 4. Leave the other parameters along and be sure the transmission box is selected. Note: in the windows version the background counts, iterations and regularization parameters aren’t functional. So, no adjustment in these values is necessary.

Select OK which will result in 4 tomograms that have been reconstructed with 0, 4, 8, and 12 as the center of rotation. Be patient, depending on the computer it can take some time to reconstruct all the slices. 8. An easy way to view the images and assess which is the sharpest is to stack all the images and then scroll through them quickly. To do this go to Image on the Fiji menu bar and choose Stacks, Images to Stacks. Now use the mouse roller button to scroll through the reconstructed images and choose the center the sharpest image and note the center. 9. Now choose the image that looks best and repeat the process, but this time with a smaller low and high range and smaller step size until the sharpest image is achieved. Note: To repeat the centering process requires that all the sinogram windows be closed and steps 4-7 repeated.

Figure 2

Figure 3

10. Once the best center has been found, repeat steps 1-7 but this time enter the same low and high center of rotation values with a step of 1 (which doesn’t really matter since there is no step between the same values) and uncheck the transmission box. Press OK 11. The result will be the reconstruction of all fluorescent channels. This will take some time. When completed a new tomogram window will pop up displaying all of the reconstructed slices as a stack. You can scroll through the stack with the roller ball on the mouse or by the slide at the bottom of the window. Save this file as a .TIFF. 12. To decouple the stack of images and produce individual images for each of the elements go to Image on the Fiji menu bar, select Stacks and then Stacks to Images (Figure 4). The result will be new windows each containing a tomogram showing the distribution of the individual element. The header on the window will indicate which element the tomogram is showing. These show up as numbers only and correspond to the order of ROI’s predefined in the XY or XE mapping program. 13. Save each of the individual element specific tomograms as .TIF files for later processing Image Processing 1. Open the file you wish to process using the open file options on the Fiji menu or by simply dragging the file to the Fiji toolbar from the file folder. 2. Some things that you may want to do to the images are to enhance the contrast, reduce the background, increase the brightness, smooth the image, etc. Most of these can be

Figure 4

done through Fiji’s Image menu (Figure 4). Smoothing is found under the Process menu option. NOTE: Images produced in Fiji are in 32-bit format which is not support by all graphic programs. Newer versions of MS photoshop and possibly 32-bit versions of MS paint can deal with these images. The point is to maintain 32-bit format while working on the file in Fiji to retain the desired imaged resolution. To do so, make all adjustments to the grayscale image only (i.e. don’t convert to 8-bit RGB) and save it as a .TIF and then color the image in another 32-bit compatible program.