Most of our EPMA work is quantitative thin film analysis on Si substrates and there is really nothing that difficult about it, but it does help to run through the steps I think.
The biggest difference in the acquisition for thin film characterization is that one needs to acquire x-ray intensities at multiple electron beam energies for *both* the standards and unknowns.
Strictly speaking that isn't always necessary, but we find it provides the most accurate and reproducible results. In fact, because we will utilize all the beam energies when correcting for the matrix effects and the thin film sample geometry, we can often obtain a more "robust" analysis than using only one electron beam energy. Of course the usual caveats apply, especially surface contamination, native oxide and/or hydrocarbon layers, and insulating and/or beam sensitive films.
And don't even think that you can avoid correcting for geometry- unless you are, one, operating at extremely low overvoltages, i.e., shallow probe, two, using a very, very inclined sample to the beam, or three, your sample is thick enough that your electrons do not penetrate to the substrate, i.e., your sample is not really a thin film!
Basically we will be collecting x-ray intensities for our thin film elements, the substrate elements (to help with our thickness calculations) and also any contaminate elements such as oxygen. One time when we started seeing low totals in our thin films (yes, we also get useful totals in thin film analysis), we discovered that the samples were getting contaminated with Se from a previous deposition!
But in practice this method works very well, exactly because most of the beam does penetrate to the substrate depositing most of the energy in pure Si! Note therefore, as general a "rule of thumb", if you detect any substrate x-rays you *must* correct for a thin film geometry in your thin samples. But remember, because we are working with thin films, we are working with fewer atoms (50 nm = 500 atoms thick!), therefore you should treat thin film analyses as you would any trace element measurements with all necessary attention paid to the background characterization.
Ok, so the first step is to create a sample setup containing all the elements you need. Once you are tuned up and everything looks good (you did remember to run a careful wavescan to check for interfering off-peaks didn't you?), you will create three new sample setups based on the first and name them for the beam energies that they will have, for example 10 keV, 15 keV and 20 keV.
Also consider using multi-point backgrounds for highly variable thin film compositions where the background positions utilized for the background correction are "dynamic" and can change "on the fly" as discussed here:
http://smf.probesoftware.com/index.php?topic=56.msg218#msg218
Of course the actual multiple voltages you utilize depends on the thickness of the thin film and the elements and x-ray lines you are using. Thinner films require lower voltages and lower energy x-ray lines sometimes. It is suggested that you model the film's likely chemistry and thickness and decide which voltages will produce the largest change in intensity with a change in electron beam energy.
(https://smf.probesoftware.com/oldpics/i41.tinypic.com/333fw9v.jpg)
What we are after is a variation in the thin film element x-ray intensities as a function of beam energy as seen here:
(https://smf.probesoftware.com/oldpics/i44.tinypic.com/343gs9w.jpg)
The lines are the model (PAP) and the symbols are the EPMA measurements. From these measurements we can obtain both the composition and thickness of the film. Note that the signal from the substrate increases as the beam energy is increased.
More caveats, it is very difficult to measure a thin film containing an element that is also in the substrate. In the above case oxygen is present in the film, so we would not want to try measuring a thin film for oxygen on a substrate of, for example MgO. Or say, SiGe films on Si, though it *can* be done if the thickness is already known, see this paper for more details:
http://epmalab.uoregon.edu/pdfs/Determination%20of%20Ni-Si%20Ultra-thin%20Films,%20Phung,%20et.%20al.,%202008.pdf
Next here's an example using La lines for mid-Z elements and this film is even thinner than the previous one. Also note the oxygen native oxide intensity increases as the voltage is lowered.
(https://smf.probesoftware.com/oldpics/i43.tinypic.com/25s3g4k.jpg)
So now back to our acquisition.
After creating and naming our sample setups for the beam energies we decided on, we then change the operating voltage on each sample (from the Analyze! window is best because then the instrument doesn't actually have to change beam energies) using the Conditions button.
Next we want to save these samples as sample setups that can be referenced for digitizing our samples. This is easily done by selecting all three samples and clicking the Add To Setup button in the Analyze! window as seen here:
(https://smf.probesoftware.com/oldpics/i40.tinypic.com/6ny48z.jpg)
Now we are ready to digitize our unknowns (assuming the standard positions have already been loaded into the Automate! window) using the Digitize window.
I generally digitize between 6 and 10 points per thin film because they are almost perfectly uniform when deposited using thermal evaporation, but films that are deposited by laser ablation can vary in both thickness *and* composition, so be careful. You do not want to average together points during the analysis that have different thickness and/or chemistry.
Once your points are digitized (and I usually check that I get all the little chips mounted on a 25 mm Al stub by using the StageMap window as seen here):
(https://smf.probesoftware.com/oldpics/i43.tinypic.com/24xm2ya.jpg)
You can next select all your standards and unknown position samples in the Automate! window:
(https://smf.probesoftware.com/oldpics/i41.tinypic.com/11iq2p1.jpg)
select the three sample setups for your three beam energies, and note the "automation basis" is now Use Digitized Multiple Setups.
And away you go!
Thank you for this, it is a clear guide should I ever have to do this type of analysis. ;D
If you want to know why multi-voltage EPMA (WDS) is the best tool for the job of characterizing the composition (and thickness) of thin films deposited on substrates, a look at the following images should suffice:
(https://smf.probesoftware.com/oldpics/i39.tinypic.com/2w6vyhk.jpg)
and this:
(https://smf.probesoftware.com/oldpics/i42.tinypic.com/jr4ho1.jpg)
The full PPT presentation is here if you are interested:
http://epmalab.uoregon.edu/publ/Nano-2011_TEPNA.ppt
Usually, the measured k-ratios (at multiple beam energies), for thin film samples are then exported from Probe for EPMA and corrected for sample geometry using an external application such as STrataGEM or GMRFilm using this window accessible from the Output menu:
(https://smf.probesoftware.com/oldpics/i62.tinypic.com/17wswl.jpg)
This is the easiest and most accurate solution for determination of thin films where both the thickness and composition are unknown, but it should be noted that for thin films that are "unsupported" or deposited on a low Z substrate (does Si qualify as "low" Z? It depends... on the physics details!) and where the film thickness (and density) are fairly well known, one can utilize a variation on the particle correction procedures in Probe for EPMA that are described here for particles:
http://smf.probesoftware.com/index.php?topic=281.0
Normally, in thin film analysis where the thickness is unknown we measure the k-ratios at multiple beam energies including the intensities emitted from the substrate (usually Si Ka and O ka for Si wafer depositions), in order to help constrain the thin film thickness during the physics interation and allow solving for composition and thickness simultaneously.
However, in the case of a known thickness we could utilize a single beam energy, although it may help estimate accuracy by utilizing several beam voltages. And because the substrate is not included in this "particle" calculation we'll want to want an unsupported film or low Z substrate to minimize fluorescence issues and *not* measure the substrate element. To see what I mean let's start with a roughly 50 nm Mo,W oxide thin film deposited on a Si wafer and analyze for Mo, W, O (and also Si just to see why we do not want to include the substrate emissions) as seen here without any correction for thin film geometry at a beam energy of 10 keV:
Un 6 WMo oxide-1, Results in Elemental Weight Percents
ELEM: Mo W O Si
BGDS: LIN LIN EXP HIGH
TIME: 60.00 60.00 60.00 60.00
BEAM: 29.94 29.94 29.94 29.94
ELEM: Mo W O Si SUM
162 2.924 14.522 12.619 70.344 100.409
163 2.902 14.240 12.310 69.995 99.447
164 2.926 14.447 12.549 70.279 100.202
165 2.926 14.674 12.355 70.107 100.062
166 2.912 14.351 12.377 70.229 99.869
167 2.911 14.180 12.349 70.226 99.666
168 2.910 14.269 12.367 70.130 99.675
169 2.889 14.434 12.421 70.069 99.813
170 2.973 14.324 12.479 70.033 99.808
171 2.846 14.386 12.244 70.265 99.742
172 2.777 14.141 12.222 70.068 99.208
173 2.926 14.266 12.043 70.447 99.682
174 2.849 14.247 12.173 70.423 99.692
175 2.911 14.408 12.391 70.307 100.017
176 2.894 13.969 12.224 70.237 99.324
AVER: 2.898 14.324 12.342 70.210 99.774
SDEV: .046 .169 .148 .140 .317
SERR: .012 .044 .038 .036
%RSD: 1.57 1.18 1.20 .20
STDS: 542 574 913 514
The above analysis is a great example of "spurious accuracy" in that the seemingly good total is completely wrong. Why? Because the Si is not in the film and including it in the matrix correction is not a physical model. We can check this because we know the film chemistry should be approximately (Mo,W)O3 or 1 to 3 cation to oxygen and the following atomic percent output is clearly not even close as seen here:
Un 6 WMo oxide-1, Results in Atomic Percents
ELEM: Mo W O Si SUM
162 .896 2.321 23.179 73.604 100.000
163 .898 2.299 22.836 73.968 100.000
164 .898 2.314 23.098 73.690 100.000
165 .903 2.362 22.855 73.880 100.000
166 .897 2.308 22.869 73.926 100.000
167 .898 2.282 22.837 73.983 100.000
168 .898 2.298 22.883 73.922 100.000
169 .891 2.323 22.969 73.817 100.000
170 .916 2.303 23.059 73.722 100.000
171 .879 2.319 22.675 74.128 100.000
172 .860 2.286 22.704 74.150 100.000
173 .905 2.303 22.342 74.450 100.000
174 .880 2.296 22.540 74.285 100.000
175 .896 2.314 22.869 73.921 100.000
176 .895 2.254 22.664 74.187 100.000
AVER: .894 2.305 22.825 73.975 100.000
SDEV: .013 .024 .217 .231 .000
SERR: .003 .006 .056 .060
%RSD: 1.45 1.03 .95 .31
The same sample but calculated at 20 keV, gives yet another set of different but also wrong values:
Un 8 WMo oxide-1, Results in Elemental Weight Percents
ELEM: Mo W O Si
BGDS: LIN LIN EXP HIGH
TIME: 60.00 60.00 60.00 60.00
BEAM: 29.76 29.76 29.76 29.76
ELEM: Mo W O Si SUM
192 1.065 3.575 10.340 91.192 106.172
193 1.066 3.535 10.316 91.163 106.078
194 1.060 3.653 10.295 91.339 106.348
195 1.082 3.625 10.158 91.253 106.117
196 1.023 3.609 10.208 91.149 105.989
197 1.056 3.597 10.172 91.350 106.176
198 1.077 3.552 10.176 91.511 106.317
199 1.070 3.571 10.182 91.333 106.157
200 1.043 3.709 10.122 90.990 105.863
201 1.047 3.642 10.300 91.374 106.362
202 1.008 3.536 10.090 91.469 106.104
203 1.040 3.607 10.006 91.031 105.683
204 1.073 3.572 10.044 91.624 106.313
205 1.010 3.582 10.141 91.408 106.141
206 1.007 3.613 10.193 91.307 106.120
AVER: 1.049 3.598 10.183 91.299 106.129
SDEV: .026 .047 .098 .174 .183
SERR: .007 .012 .025 .045
%RSD: 2.46 1.30 .96 .19
STDS: 542 574 913 514
STKF: .9915 .9978 .1970 1.0000
STCT: 415.35 144.62 122.91 291.32
UNKF: .0059 .0407 .0229 .8755
UNCT: 2.46 5.90 14.26 255.06
UNBG: .22 .48 1.26 .24
ZCOR: 1.7824 .8833 4.4546 1.0428
KRAW: .0059 .0408 .1160 .8755
PKBG: 12.01 13.32 12.30 1078.89
INT%: ---- ---- -.18 ----
APF: ---- ---- .999 ----
Un 8 WMo oxide-1, Results in Atomic Percents
ELEM: Mo W O Si SUM
192 .283 .496 16.471 82.751 100.000
193 .283 .490 16.443 82.783 100.000
194 .281 .506 16.387 82.825 100.000
195 .288 .504 16.217 82.992 100.000
196 .272 .502 16.302 82.924 100.000
197 .281 .499 16.223 82.997 100.000
198 .286 .492 16.205 83.017 100.000
199 .285 .496 16.239 82.980 100.000
200 .278 .517 16.207 82.998 100.000
201 .278 .504 16.389 82.829 100.000
202 .268 .491 16.099 83.142 100.000
203 .278 .503 16.048 83.171 100.000
204 .285 .496 16.012 83.208 100.000
205 .269 .497 16.175 83.059 100.000
206 .268 .502 16.259 82.972 100.000
AVER: .279 .500 16.245 82.976 100.000
SDEV: .007 .007 .136 .137 .000
SERR: .002 .002 .035 .035
%RSD: 2.41 1.38 .84 .17
Clearly still not close to 3:1, so let's "disable quant" on the Si Ka intensities and recalculate (again still without the thin film correction) and see if that improves things at all using the 10 keV intensities:
Un 6 WMo oxide-1, Results in Elemental Weight Percents
ELEM: Mo W O Si
BGDS: LIN LIN EXP HIGH
TIME: 60.00 60.00 60.00 ---
BEAM: 29.94 29.94 29.94 ---
ELEM: Mo W O Si-D SUM
162 2.620 13.956 11.196 --- 27.772
163 2.598 13.679 10.929 --- 27.207
164 2.621 13.883 11.137 --- 27.641
165 2.621 14.080 10.992 --- 27.693
166 2.608 13.784 10.989 --- 27.382
167 2.606 13.626 10.958 --- 27.190
168 2.605 13.709 10.978 --- 27.292
169 2.588 13.864 11.025 --- 27.477
170 2.663 13.767 11.084 --- 27.514
171 2.549 13.808 10.870 --- 27.227
172 2.487 13.581 10.826 --- 26.895
173 2.618 13.687 10.717 --- 27.022
174 2.550 13.675 10.805 --- 27.031
175 2.607 13.837 11.003 --- 27.447
176 2.590 13.425 10.842 --- 26.857
AVER: 2.595 13.757 10.957 --- 27.310
SDEV: .041 .160 .129 --- .285
SERR: .011 .041 .033 ---
%RSD: 1.58 1.16 1.18 ---
STDS: 542 574 913 ---
STKF: .9901 .9971 .3323 ---
STCT: 162.36 73.01 135.61 ---
UNKF: .0206 .1188 .0605 ---
UNCT: 3.37 8.70 24.70 ---
UNBG: .18 .25 1.48 ---
ZCOR: 1.2613 1.1580 1.8102 ---
KRAW: .0208 .1192 .1821 ---
PKBG: 19.91 36.16 17.74 ---
INT%: ---- ---- -.60 ---
APF: ---- ---- .980 ---
Un 6 WMo oxide-1, Results in Atomic Percents
ELEM: Mo W O Si-D SUM
162 3.401 9.453 87.146 --- 100.000
163 3.452 9.484 87.064 --- 100.000
164 3.420 9.452 87.128 --- 100.000
165 3.455 9.683 86.862 --- 100.000
166 3.445 9.503 87.052 --- 100.000
167 3.455 9.427 87.118 --- 100.000
168 3.447 9.465 87.088 --- 100.000
169 3.408 9.527 87.064 --- 100.000
170 3.489 9.414 87.097 --- 100.000
171 3.401 9.616 86.983 --- 100.000
172 3.339 9.514 87.147 --- 100.000
173 3.537 9.649 86.814 --- 100.000
174 3.424 9.582 86.994 --- 100.000
175 3.439 9.525 87.036 --- 100.000
176 3.471 9.390 87.139 --- 100.000
AVER: 3.439 9.512 87.049 --- 100.000
SDEV: .045 .087 .100 --- .000
SERR: .012 .022 .026 ---
%RSD: 1.31 .91 .11 ---
The total is very low because we have not yet turned on the thin film correction, and the cation to oxide ratios are better than before, but still not 1:3. So now let's turn on the thin film correction as seen here:
(https://smf.probesoftware.com/oldpics/i59.tinypic.com/5cbmfo.jpg)
where I have selected the "Thin Film" option, entered 50 nm for the thickness (0.05 um) and 5 for the density (I have no idea what it might actually be from a spin cast deposition subsequently baked)...
and output the compositions for all three beam energies, first at 10 keV:
Un 6 WMo oxide-1, Results in Atomic Percents
ELEM: Mo W O Si-D SUM
162 4.497 14.884 80.619 --- 100.000
163 4.570 14.949 80.481 --- 100.000
164 4.524 14.887 80.589 --- 100.000
165 4.574 15.296 80.130 --- 100.000
166 4.560 14.980 80.459 --- 100.000
167 4.574 14.853 80.574 --- 100.000
168 4.563 14.915 80.522 --- 100.000
169 4.508 15.014 80.477 --- 100.000
170 4.622 14.838 80.540 --- 100.000
171 4.499 15.165 80.336 --- 100.000
172 4.410 14.974 80.616 --- 100.000
173 4.690 15.256 80.054 --- 100.000
174 4.531 15.112 80.357 --- 100.000
175 4.551 15.018 80.431 --- 100.000
176 4.596 14.791 80.613 --- 100.000
AVER: 4.551 14.996 80.453 --- 100.000
SDEV: .064 .152 .172 --- .000
SERR: .016 .039 .044 ---
%RSD: 1.40 1.01 .21 ---
and at 15 keV:
Un 7 WMo oxide-1, Results in Atomic Percents
ELEM: Mo W O Si-D SUM
177 4.512 14.040 81.448 --- 100.000
178 4.390 14.299 81.311 --- 100.000
179 4.513 14.393 81.094 --- 100.000
180 4.310 14.576 81.114 --- 100.000
181 4.533 14.327 81.141 --- 100.000
182 4.444 14.485 81.071 --- 100.000
183 4.507 14.191 81.302 --- 100.000
184 4.479 14.533 80.988 --- 100.000
185 4.401 14.717 80.882 --- 100.000
186 4.426 14.360 81.214 --- 100.000
187 4.504 14.552 80.944 --- 100.000
188 4.509 14.381 81.111 --- 100.000
189 4.504 14.566 80.931 --- 100.000
190 4.521 14.455 81.025 --- 100.000
191 4.526 14.298 81.176 --- 100.000
AVER: 4.472 14.411 81.117 --- 100.000
SDEV: .064 .171 .156 --- .000
SERR: .017 .044 .040 ---
%RSD: 1.44 1.19 .19 ---
and finally at 20 keV:
Un 8 WMo oxide-1, Results in Atomic Percents
ELEM: Mo W O Si-D SUM
192 4.379 13.480 82.141 --- 100.000
193 4.396 13.376 82.228 --- 100.000
194 4.362 13.794 81.844 --- 100.000
195 4.502 13.856 81.642 --- 100.000
196 4.253 13.773 81.974 --- 100.000
197 4.401 13.762 81.837 --- 100.000
198 4.491 13.598 81.911 --- 100.000
199 4.458 13.655 81.887 --- 100.000
200 4.347 14.198 81.455 --- 100.000
201 4.310 13.760 81.930 --- 100.000
202 4.246 13.690 82.063 --- 100.000
203 4.392 14.006 81.602 --- 100.000
204 4.520 13.830 81.650 --- 100.000
205 4.231 13.780 81.989 --- 100.000
206 4.195 13.817 81.988 --- 100.000
AVER: 4.365 13.758 81.876 --- 100.000
SDEV: .103 .196 .212 --- .000
SERR: .027 .051 .055 ---
%RSD: 2.35 1.43 .26 ---
and we can see that the atomic rations for Mo + W to O is roughly 1:4 for all three beam energies which gives us some confidence that we might be doing something right after all!
But why is our cation to oxygen ratio still too high? It really ought to be 1:3, not 1:4. Where could that extra oxygen be coming from?
Well one obvious explanation is that the Si substrate is backscattering and/or fluorescing oxygen in the film from the substrate to some small but significant degree (which is why we really should be using carbon planchet substrate or a physics model such as STrataGEM or GMRFilm which includes the substrate effects)... remember, our oxide film is only some 50 nm thick and it wouldn't take much to produce some significant additional oxygen from backscatter electrons or fluorescence from the substrate.
The other obvious question is: is this substrate really oxygen free? We know that Si wafers continue to grow their native oxide layer over time and when I asked the student: "did you etch the Si wafer prior to deposition?", the answer was, sadly, no...
Consider a 50 nm film on top of a 5 nm or so thick native oxide layer. That is an additional 10% oxygen!
So all in all a pretty good result considering that we shouldn't really be doing it this way at all. If anyone else has measurements on a better specimen please feel free to post your examples.
Another way to evaluate the "crude but sometimes effective" thin film correction in PFE is to directly compare its results to a full treatment with multi-voltage analysis (MVA).
Here are the results of a BiSnTi thin film sample on Si calculated with STrataGEM using intensities measured at 7, 11 and 15 keV:
(https://smf.probesoftware.com/oldpics/i57.tinypic.com/5u0qi9.jpg)
(https://smf.probesoftware.com/oldpics/i61.tinypic.com/29m69uf.jpg)
Note the normalized weight fractions in the column labeled "Weight". Now compare those to the results using the PFE thin film correction at the three voltages. First at 7 keV:
Un 35 alloy_1b, Results in Normalized Elemental Weight Percents (Particle Corrections)
ELEM: Si-D O Se Ti Bi Sn SUM
241 --- .538 57.815 10.546 23.942 7.159 100.000
242 --- .560 58.251 10.426 23.688 7.074 100.000
243 --- .544 57.984 10.434 24.106 6.932 100.000
244 --- .475 58.481 10.797 23.157 7.090 100.000
245 --- .506 58.343 10.600 23.516 7.035 100.000
AVER: --- .525 58.175 10.561 23.682 7.058 100.000
SDEV: --- .034 .271 .151 .371 .083 .000
SERR: --- .015 .121 .068 .166 .037
%RSD: --- 6.49 .47 1.43 1.57 1.18
Now at 11 keV:
Un 36 alloy_1b, Results in Normalized Elemental Weight Percents (Particle Corrections)
ELEM: Si-D O Se Ti Bi Sn SUM
246 --- .746 59.812 9.263 23.488 6.692 100.000
247 --- .691 60.084 9.475 23.068 6.683 100.000
248 --- .671 59.548 9.356 23.792 6.633 100.000
249 --- .741 60.326 9.351 22.803 6.779 100.000
250 --- .778 60.217 9.501 22.343 7.160 100.000
AVER: --- .725 59.997 9.389 23.099 6.789 100.000
SDEV: --- .044 .316 .098 .568 .214 .000
SERR: --- .020 .141 .044 .254 .096
%RSD: --- 6.04 .53 1.05 2.46 3.15
Finally at 15 keV:
Un 37 alloy_1b, Results in Normalized Elemental Weight Percents (Particle Corrections)
ELEM: Si-D O Se Ti Bi Sn SUM
251 --- .943 60.638 9.221 22.722 6.477 100.000
252 --- .657 60.977 9.153 22.544 6.669 100.000
253 --- .990 60.789 9.139 22.432 6.650 100.000
254 --- .704 60.493 9.246 22.786 6.771 100.000
255 --- .926 60.824 9.135 22.427 6.688 100.000
AVER: --- .844 60.744 9.179 22.582 6.651 100.000
SDEV: --- .152 .185 .051 .165 .108 .000
SERR: --- .068 .083 .023 .074 .048
%RSD: --- 18.02 .30 .56 .73 1.62
As you can see even the crude "unsupported thin film" model compares well to the full substrate MVA physics model- at least in this one example!
The most recent versions of StrataGEM now require a $End statement at the end of the import file created by Probe for EPMA. And no, this change made by JF Thiot is *not* backward compatible with earlier versions of StrataGEM (don't ask me why!).
Anyway, here is the updated documentation in the Probe for EPMA help file. Basically you need to add this new keyword to the [software] section of your Probewin.ini file if you are using StrataGEM v. 5 or earlier (and maybe even some early releases of version 6 as well according to JF Thiot!).
StrataGEMVersion=6
This flag is used to add a "$End" statement at the end of the StrataGEM import file. Beginning with version 6 of StrataGEM a $End statement is required at the end of the import file.
However, versions of StrataGEM prior to version 6 *cannot* have a $End statement at the end of the StrataGEM import file according to JF Thiot. Whatever happened to backward compatibility?
Anyway, the default StrataGEM version is 6, so Probe for EPMA will add a $End statement to the end of the import file automatically, unless the StrataGEMVersion keyword in the Probewin.ini file [software] section is set to 4 or 5. Which means that users with older versions of StrataGEM will have to edit their Probewin.ini files manually.
Dear John,
recently I started to work with thin electron-transparent samples prepared using FIB SEM. Such samples are mounted on copper grid, no substrate used. It can see some potential in such analysis, particularly for high spatial resolution applications at "normal" and high voltages. Quantitative WDS of thin films is less involved compared to the bulk, but is tricky as calibration and correction philosophies have to be modified. While working on wedged samples I could observe how ZAF-based corrections stop working with sample thickness reduction. Surprisingly thin film corrections are not working properly either as absorption coefficients are rather high. I am thinking of using Cliff - Lorimer approach and wonder what is your take on it. Would you recommend to use calibration curve routines?
Cheers,
Sergei
Quote from: smatveev on July 17, 2015, 01:49:48 AM
Dear John,
recently I started to work with thin electron-transparent samples prepared using FIB SEM. Such samples are mounted on copper grid, no substrate used. It can see some potential in such analysis, particularly for high spatial resolution applications at "normal" and high voltages. Quantitative WDS of thin films is less involved compared to the bulk, but is tricky as calibration and correction philosophies have to be modified. While working on wedged samples I could observe how ZAF-based corrections stop working with sample thickness reduction. Surprisingly thin film corrections are not working properly either as absorption coefficients are rather high. I am thinking of using Cliff - Lorimer approach and wonder what is your take on it. Would you recommend to use calibration curve routines?
Cheers,
Sergei
Hi Sergei,
With thin film samples (substrate or no substrate) one must correct for the fact that the electrons do not come to rest in the material of interest (relative to a standard). I have found that the thin film correction in CalcZAF gives excellent results especially when the substrate is missing (TEM mounts) or is a low Z substrate where the substrate element(s) does not fluoresce any thin film elements. Otherwise I use STRATAGem* routinely. About 60% of the work in our lab is thin films on substrates.
This post describes correcting for small particle geometries in CalcZAF, but if you use the first option (thin film without substrate), you should get good results:
http://smf.probesoftware.com/index.php?topic=281.msg2735#msg2735
* © Copyright 1993-2016 SAMx
Hi All
I do not have any experience with analyzing thin films, but got a request to do some analysis on CoFeB layers. I just printed out John D.'s directions above, but will take me time to understand and do tests to see if I can actually do this on my new microprobe. Somewhat time consuming. Is there anyone experienced in doing thin films looking for some analysis work and to help out a grad student? This is for grad student in Engineering in Brown Univ. I can send you the details.
Thanks Joe
In case you do decide to try your hand at thin film analysis- you might want to grab the free BadgerFilm app here:
https://smf.probesoftware.com/index.php?board=37.0
It takes the same output file of stds and k-ratios from Probe for EPMA and is easy to use (and doesn't cost thousands of dollars!).
Quote from: Joe Boesenberg on August 26, 2025, 01:21:58 PMIs there anyone experienced in doing thin films looking for some analysis work and to help out a grad student? This is for grad student in Engineering in Brown Univ. I can send you the details.
Julie Chouinard at Univ Oregon is very experienced in analyzing thin films quantitatively:
https://camcor.uoregon.edu/facilities-and-instrumentation/microanalytical
John
Thanks for both suggestions. Joe