UC Berkeley, Geology & Geophysics Electron Microprobe
Summary of Specification Testing :
The following is a brief summary of the testing performed on the Cameca microprobe during the past 2 months. Please refer to the specification listing for a complete description of the requirements. For details on the specification test results please contact John Donovan.
1.
a) Current stability : Measured 0.068% per hour (specification was 0.05%) in 12 hours and 0.526% per hour (specification was 0.5%) in 24 hours. This was before an air conditioner was installed. The room had been oppressively warm after running overnight so this result should now be much improved. I will run this test again but I am satisfied with these results in any case.
b) Automatic gun bias : This works correctly as determined by EDS Duane Hunt limit tests.
c) Current range and accuracy : The current adjustment range is fine, however I am not able to independently confirm the accuracy of the beam current.
d) Filament life : Unsatisfactory, please see accompanying letter.
2.
a) Voltage range : This is satisfactory
b) Voltage accuracy : Measured 15.05 KeV using EDS Duane Hunt limit test. 100 nA on Mo sample at 15 KeV.
c) High voltage stability : Accept Cameca France measurements performed by Bernard Vautier.
d) Beam diameter : Beam diameter is unsatisfactory. Using SE I would estimate that the beam diameter at 50 nA, 15 KeV is closer to 2 microns (specification is 0.5 microns).
e) Stray beam : Using a 100 microns aperture I measure 4245.7 cps of Mo ka on the metal at 20 KeV and 100 nA and 0.2 cps in the center of the aperture. This results in a k-ratio of 0.000047 (specification is 0.0001). Satisfactory.
f) Beam diameter adjustment : The "size" command will adjust from 1 to 250 microns (specification is 1 to 100).
g) Beam current change and image shift during spectrometer full range move : The measured beam current change was 0.06 % and about 0.2 u vibration (the specification required .5% beam current stability and 1 micron image stability) while all 5 spectrometers moved.
h) Beam focus and position stability during condenser lens adjustment : No change in beam position was noted from 1 to 10 nA.
i) Final apertures : Apertures are selectable and adjustable
j) Beam current variation measured on carbon and Fe : The variation is approximately 0.06% (specification was 0.5%). However this good result was overshadowed by the observation that the beam current is somewhat affected by the stage moving, possibly due to some shielding problem on the stepper motors. The variation of beam current between one stage position and another was measured as 0.3%. This problem was verified by the Cameca engineer Bernard Vautier. There is no specific specification for this result, however, I would hope that Cameca will try in good faith to determine the cause of this beam current shift due to stage position and fix it.
k) Carbon contamination rate : I would just like to note that although the Cameca contamination rate is within specification, I find it disappointing that my 30 year old SEMQ microprobe has a much lower carbon contamination rate because of a simple chilled Freon baffle over the diffusion pump. The use of a chilled baffle such as this eliminates the need for LN2 traps and oxygen leak devices. The measured carbon contamination rate on the Cameca is 0.04% per min (specification is 0.16%) without any anti-contamination and 0.017% per min (specification is 0.01%) with only the oxygen leak. The use of the LN2 trap improves it by another factor of two.
l) Absolute accuracy : These measurements have not been made yet due to delays in obtaining suitable magnification standards. They will be run before the final acceptance period.
m) Column window justification : I agree with Cameca that the benefits of a column window (improved vacuum, live adjustment of crystals, etc.) outweigh the decrease in light element intensities. The drop in intensity is only several percent for O ka and C ka, although about 15% - 30% for N ka.
n) EDS aperture turret : This device is excellent. After installing an 8 u Be window in one of the positions, we find that it is possible to use the EDS ultra thin window system even with the light optics on. If light element EDS is required we can simply move the Be window aside and turn on the light. Cameca did supply schematic drawings as specified.
Note that the original turret position on the shaft was too close to the snout of the Oxford EDS detector, requiring that I drill out the set screws (they were staked in place) and move it slightly away too avoid an electrical short.
3.
a) Five WDS spectrometers : All spectrometers were shipped as specified and 4 blank crystal holders were received. One has been used already to mount one of our LDE multilayers from LBL.
b) High speed spectrometer motors : Measured 14 seconds end to end including the 4 second delay after the command is given (specification was 20 seconds).
c) Simultaneous use of WDS, EDS light optics : No problem.
d) Crystal flip in any position : This is an interesting specification. Originally I had asked Cameca this question and there must have been some miscommunication because I thought they said "yes, the crystals can flip in any position". That is why this specification exists. In fact, the crystals must be at position 78000 (or higher) to flip. However, they can start the flip procedure from any position by moving automatically to the flip position from any position. Given the vastly superior engineering of the Cameca spectrometers (the best part of the instrument in my opinion) I am willing to cede this point to Cameca.
e) Spectrometer reproducibility : This is a tough test and the Cameca spectrometers really shined. Conditions were 40 seconds count time, 15 Kev 30 nA. All values in percent change in total counts on each of two peaks at one-half the peak intensity (average of 10 measurements).
| spec # | 1 | 2 | 3 | 4 | 5 |
| w/o flip | .50/.38 | .56/.31 | .82/.91 | .59/.72 | .68/.60 |
| w/ flip | 1.15/1.31 | 1.48/1.25 | .94/1.3 | 4.05*/1.29 | 2.20/2.21 |
* ADP crystal has an extremely sharp peak
The specification requires reproducibility within .6% w/o crystal flip and within 2% with a crystal flip. Except for spectrometer #4 on the ADP all spectrometers performed perfectly.
f) Simultaneous k-ratios : Another tough one. The measured ratio of a low energy x-ray on two different compositions, should be the same for all spectrometers the intensities are measured on. This test determines if the crystals are symmetrically diffracting, the takeoff angle is correct and the stage is level. Conditions were measured on pure copper and Cu80Au20 using Cu La at 10 nA., and 15 KeV. Average of 5 measurements of 30 sec each. The measurements were performed on the light element spectrometers (#1, #4, #5) which are spaced approximately at the corners of an equilateral triangle around the column. All values in percent change in k-ratio value.
| spec # | 1 | 2 | 3 | 4 | 5 |
| test 1 | -.27 | -.47 | -.13 | ||
| test 2 | .49 | -.42 | -.12 |
The specification requires the values to be within 1%.
g) Symmetrical crystal diffraction : Because the Cameca analyzer crystals are permanently mounted on a holder that can only be mounted one way, it was not possible to perform this test as specified. However, the above simultaneous k-ratio test can easily determine if the crystals are not diffracting symmetrically and it appears that they are.
h) Spectrometer resolution : High resolution spectrometers are generally to be avoided for quantitative analysis, except in the case of peak overlaps. I measured an overlap k-ratio of 0.0040 and 0.0049 (about 1/2 percent) of V ka on TiO2. The specification was 0.005. This is an excellent result given that the Cameca spectrometers have no spectrometer slits.
i) Crystal count rates : All values in counts per second per micro amp.
| Crystal | CPS P/B | 1 | 2 | 3 | 4 | 5 | |
| W/Si | C ka | 1.4 105 120 |
6.2 104 | ||||
| N ka | 2.5 104 30 |
2.6 104 | |||||
| O ka | 1.8 105 120 |
2.7 105 111 |
|||||
| TAP | Mg ka | 2.8 106 1150 |
2.2 106 943 |
2.8 106 1233 |
2.8 106 1199 |
||
| Al ka | 3.5 106 1050 |
2.7 106 2008 |
3.5 106 1875 |
3.4 106 2008 |
|||
| Si ka | 3.8 106 900 |
3.0 106 1385 |
3.9 106 1725 |
3.8 106 1557 |
|||
| PET | Ti ka | 2.6 106 600 |
1.3 106 420 |
1.8 106 385 |
1.5 106 388 |
1.2 106 400 |
|
| Cr ka | 2.8 106 300 |
||||||
| Mn ka | 4.0 106 350 |
||||||
| LiF | Ti ka | 6.0 105 1000 |
4.0 105 1160 |
4.1 105 1099 |
|||
| Fe ka | 1.5 106 550 |
||||||
| Cu ka | 2.0 106 340 |
Not all elements were tested, however the crystals performed reasonably well in all cases measured. The PET P/B values are noted to be significantly lower than specified although not bad.
j) Spectrometer peak at identical Z stage position : This test was performed on TAP, PET, PET, TAP and TAP (spectrometers 1, 2, 3, etc.). The peak intensities were measured at 1 u increments counting 20 seconds and on two tests the spectrometers peak intensities were within 1 micron stage position for all spectrometers at the same time.
k) specification was not tested
4.
a) SE image resolution : Under SEM conditions (20 KeV, low current, manual tweaking) the image showed 70 angstrom resolution. That is, several 4 X 5 photographs of Au on C showed .35 mm (as measured on the photo) objects at 50000 magnifications.
b) BSE resolution : Again under optimum conditions, the BSE image approached 150 angstroms resolution. Probably closer to 300 angstrom resolution.
c) BSE atomic contrast : Alpha/Beta brass sample easily contrasted.
d) SE/BSE TV, slow and photo scan rates : OK
e) SE/BSE image distortion : TV rate imaging at low magnification (less than 300) shows a distinct distortion of the images on the left side of the image. This is evidently due to inadequate performance of the scan coils at high frequencies. The images are distortion free (less than 1%) at slow scan rates.
f) Dedicated color TV monitors : OK
5.
a) Reflected/transmitted light optics with polarizer : OK
b) Color monitor for light optics : Excellent light optics images in all configurations
c) Optical resolution of 0.7 microns : Optical tests show a very flat field and high resolution. Some phosphor coloration is noted in the upper left corner of the Sony monitor. Focusing shows nice concentric focus effects.
d) Optical depth of field : Careful measurements show a light optics depth of field between one half and one microns.
e) No image shift observed as lower polarizer is rotated. Upper polarizer is not rotatable.
6.
a) Welch roughing pumps : OK
b) Vacuum : Vacuum (if I can believe the Cameca gauges) is excellent. Column is better than 1 x 10-4 Pa (7 x 10-7 Torr).
c) Interlock protection on power, water, vacuum : This was tested several times with the Cameca engineer present.
d) Electron gun and sample chamber isolation : OK
7.
a) Sample exchange without venting spectrometer chambers : OK
b) 2 sample holders of specified configuration : OK
c) Both sample holders to allow reflected and transmitted light : OK
d) 3 minutes sample exchange : This is more like 5 minutes but acceptable
8.
a) High speed stage motors and no sample tilt : OK, in fact the stage is so fast it's scary. Is there any way to slow it down a bit?
b) 15 mm/sec stage travel : See specification 8. a above.
c) No stage shift : OK as viewed at 10000X for 30 minutes
d) XY trackball : The Pindrys trackball is quite nice. Thank-you.
e) Anti-contamination : OK
f) Auto-focus : OK
9.
a) Documentation of instrument functions : documentation of the instrument is adequate though not impressive.
b) Documentation of communication protocols : communication protocol documentation is terrible. Fortunately, I will not be using them directly since I purchased the Xmas libraries for machine control.
The quantitative tests have been performed using my new Probe for Windows software and the results are better than I had hoped. Using my standard mount and compositions from the old probe, I can routinely reproduce major element concentrations to about .5% accuracy. The combination of the Cameca Instrument and my software (with Armstrong CITZAF) seems to be an excellent match.
Back To Department of Geology and Geophysics Home Page