Monte Carlo Simulations and Other Useful Software for

  • Slides: 28
Download presentation
Monte Carlo Simulations and Other Useful Software for Enhancing Teaching and Comprehension of Concepts

Monte Carlo Simulations and Other Useful Software for Enhancing Teaching and Comprehension of Concepts in EPMA and SEM John Fournelle Eugene Cameron Electron Microprobe Lab Department of Geology and Geophysics University of Wisconsin - Madison GSA Philly 2006

About My EPMA Class I have been teaching Geo 777, Electron Microprobe Analysis, for

About My EPMA Class I have been teaching Geo 777, Electron Microprobe Analysis, for 13 years, and the course has evolved much. It is taught once a year and enrollment has fluctuated from 3 -15, though in recent years it has been 3 -9. This is a semester long course (13 weeks) with a 2 hour class (~1. 5 hr lecture), and a 2 hour lab. EPMA Class parts: (1) a project is a significant part of the class, and involves delving into some feature of EPMA (not just analyzing thesis samples); (2) there are weekly assignments involving computation and in many cases Monte Carlo simulations; (3) there are weekly lab assignments; (4) a mid term exam has been replaced with weekly quizzes that keep the students “honest” and up to date with the reading material -- the class notes are published on the class website and need to be studied prior to their delivery and discussion each week. This gives rapid feedback on who is having problems with what concepts; (5) the final written exam is replaced with a lab practical exam (identification of unknown). Many years we have a field trip to a local EDS manufacturer’s facility. Class Projects: I use this as an opportunity for both myself and my students. One year we worked on evaluating Wards biotite as a potential EPMA standard (very promising); last year we went through the 4 light element “pseudo-crystals” (nominally 45, 60, 100 and 200Å) and derived the exact 2 d and refractive indices (K values), both of which had been incorrectly stated ; 2 years ago we learned how to run the PENELOPEPENEPMA Monte Carlo program and modelled secondary fluorescence in many materials (results shown below, center). This year we will work on 2 projects: comparing various schemes for correcting element migration (”volatile elements”) in alkali-rich glasses, and evaluating the new “standardless” EDS system on a new SEM (relative both to flat polished standards, and rough particles of those same materials). This poster highlights several Monte Carlo and other software programs that I utilize in teaching my EPMA class. I find that being able to “tweak” various settings and visualize instantaneously the effect of changing parameters gets messages across more than just reading a book or hearing me talk about it. Needless to say, all these programs (except GMRFilm and PENEPMA) run only under Windows.

Monte Carlo Programs for EPMA and SEM There are several Monte Carlo programs that

Monte Carlo Programs for EPMA and SEM There are several Monte Carlo programs that are useful for demonstrating concepts of inelastic and elastic scattering of electrons hitting samples in the SEM or electron probe. David Joy’s DOS programs have been extended, with a nice front end, by Raynald Gauvin et al with the program CASINO. Three geometries -- bulk materials , vertical boundaries (e. g. , lamallae), and horizontal (thin) films, can be modelled for scattering and x-ray generation and absorption. CASINO, freeware, available from www. gel. usherbrooke. ca/casino The next 4 slides demonstrate operation of CASINO.

Operation is simple: a geometry is set up and then each composition is enterred.

Operation is simple: a geometry is set up and then each composition is enterred. A default density needs to be modified for minerals. An accelerating voltage is set and number of trajectories to be run (and also option for number displayed) is set. The simulations here, 10, 000 trajectories, took ~40 seconds on at 2. 8 GHz computer.

The main window shows the progress of the scattering “live”. Red are backscattered electrons.

The main window shows the progress of the scattering “live”. Red are backscattered electrons. The far left pane shows a range of possible displays of both the range of electron scattering (and energy vs distance), and of x-rays generated in bulk Fe 3 O 4. A very useful plot is the “energy onion” which shows the ~concentric decay of beam electron energy (examples shown next).

How does beam (”spot”) size effect electron scattering? X-ray range? Will a Field Emission

How does beam (”spot”) size effect electron scattering? X-ray range? Will a Field Emission SEM give a smaller x-ray generation footprint compared to a W SEM/e-probe? Above are simulations of a 1 um diameter Fe 3 O 4 grain; 10 nm beam vs 0. 5 mm beam.

The x-ray generated (blue) and x-ray emitted (red) versus depth plots for Fe 3

The x-ray generated (blue) and x-ray emitted (red) versus depth plots for Fe 3 O 4. This is a useful exercise to demonstrate how low energy x-rays can be highly absorbed. I have students run this a various voltages to see the effect on the path length and resulting absorption.

Another useful Monte Carlo program is Win X-ray is free, download from www. montecarlomodeling.

Another useful Monte Carlo program is Win X-ray is free, download from www. montecarlomodeling. mcgill. ca Hendrix Demers, a doctoral student at Mc. Gill with R. Gauvin, is the main author. Win X-ray is a new program that builds off CASINO, generating EDS spectra for user-supplied compositions and conditions. It provides ways to model conditions and demonstrate aspects of x-ray behavior that may be otherwise difficult to show. (It applies currently only to bulk specimens, though the plan is that once it has been shown to provide accurate results for bulk materials, to work on simulating EDS spectra in particles, and thus provide a tool for correcting errors in current EDS measurements of particles. ) Here are shown two possible applications: The next 4 slides are about Win X-ray

What the setup windows look like

What the setup windows look like

Above, Win X-ray shows that the heights of EDS characteristic peaks do not scale

Above, Win X-ray shows that the heights of EDS characteristic peaks do not scale linearly with composition. A dummy material of equal amounts of Si, Ti and Zn (input in window shown to left) is excited by 15 ke. V. Students can produce spectra from their “favorite” minerals, varying compositions of solid solutions, and get a better “mental calibration” of how EDS spectra actually relate to compositions.

The “dueling” concepts of absorption (in sample and detection system) and overvoltage (generation) can

The “dueling” concepts of absorption (in sample and detection system) and overvoltage (generation) can be demonstrated with simulating Cu Ka and La x-rays. The comparison between 20 ke. V (below left) and 10 ke. V (below center) are fairly obvious. Then add in a litle bit of hydrocarbon contamination on the detector window (left right) -- which Win Xray is happy to let you do (as well as change the type of window) -- and more reality can be shown rather dramatically.

There was a recent article about Win X-ray in Microscopy and Microanalysis

There was a recent article about Win X-ray in Microscopy and Microanalysis

PENELOPE, a somewhat different Monte Carlo program for EPMA Another Monte Carlo program, PENELOPE

PENELOPE, a somewhat different Monte Carlo program for EPMA Another Monte Carlo program, PENELOPE (actually PENEPMA), has the added ability to model secondary fluorescence (SF). SF is a complicating phenomenon -because x-rays have no charge, they can travel tens to hundreds of times as far as electrons before “giving up the ghost”. This produces potential errors -- which are overlooked many times -- particular when looking at minor or trace element levels. In 2004, our EPMA class projects were to model various pairs of materials, where an element absent in one material would incorrectly appear to have some, due to the presence of the element next door. PENELOPE (PENEPMA) Results in following slides from 2004 EPMA Class Students Justin Gosses, Jacques Kelly, Kathy Staffier, Jeff Waters, and Caroline Webber PENEPMA (PENELOPE) was developed by F. Salvat and co-workers (U. Barcelona). For more information contact him at cesc@ecm. ub. es The next 4 slides show the results of students’ projects using PENELOPE/PENEPMA

1. Opx (Ca-free) adjacent to Cpx, simulating how Ca from adjacent CPX can make

1. Opx (Ca-free) adjacent to Cpx, simulating how Ca from adjacent CPX can make the apparent Ca content of Opx lamallae be too high. This can lead to slightly lower (2025°) temperatures when geothermometer calculations are run.

2. In 1965 Duke and Brett found 1 -4 wt% Fe in Cu grains

2. In 1965 Duke and Brett found 1 -4 wt% Fe in Cu grains in a stony meteorite. This was very difficult to explain except as SF. We modelled a 2 um Cu particle in Fe. The results below show that a 2 um grain could easily show 4 wt% Fe just from SF-- when there was really 0 wt% Fe.

3. Incorrect (too high) Fe in plagioclase: here we modelled Fo 90 (with not

3. Incorrect (too high) Fe in plagioclase: here we modelled Fo 90 (with not a large amount of Fe, but similar to the amount in a basaltic glass) next to An 80. There is an apparent (false) 0. 2 wt% Fe. O at 20 microns from the rim -- all due to SF from the continuum.

4. Incorrect (too) high Ca in olivine: same model, now looking at the SF

4. Incorrect (too) high Ca in olivine: same model, now looking at the SF effect due to Ca Ka in the An 80 being excited by x-rays produced only in the olivine. An excellent paper on this is Llovet and Galan, 2003, American Mineralogist, 88, pp. 121 -130.

Thin Film Modelling for EPMA with GMRFilm Geological EPMA generally does not require use

Thin Film Modelling for EPMA with GMRFilm Geological EPMA generally does not require use of the concept of “thin films” -however, there are cases where familiarity with the GMR Film program can provide “added value” to both students and lab directors. Here are two cases.

Case 1. What is the effect of a too light or too heavy carbon

Case 1. What is the effect of a too light or too heavy carbon coat on samples, compared to that on the standards? GMR Film provides a quick estimate. Here are two simulations for Fo 90 olivine, at 15 ke. V. Bottom line: 50 Å heavier carbon coat will yield slightly lower K-ratios (and same order wt% elements) for Si (0. 2%), Mg (0. 3%), Fe (0. 4%), but much lower for Oxygen (1. 7%). Thus the reason for coating the standards together with the unknowns if measuring oxygen.

Case 2. You have a gold-coated (for ion probing) zircon, which you would like

Case 2. You have a gold-coated (for ion probing) zircon, which you would like to have an EPMA analysis, but you don’t want to remove the Au coat, nor do you have the ability to Au coat your zircon standards. What to do? The Au coat will both retard the electron probe energy as well as absorb x-rays escaping the same (and also providing a source of secondary fluorescence). Solution? Punt, with GMR Film. Set up a theoretical zircon composition, with a guessed at Au thickness, and attempt to match the experimental K-ratios with ones calculated by the program from theory. Keep iterating until you get a close match, which then provides a best guesstimate to the zircon’s composition.

There is only one publication relating to GMR Film: “An iteration procedure to calculate

There is only one publication relating to GMR Film: “An iteration procedure to calculate film compositions and thicknesses in electronprobe microanalysis”, R. E. Waldo, in the 1988 Microbeam Analysis volume (D. E. Newbury, ed. ) [MAS annual proceedings], p. 310 -314. The software at one time was available on websites of the Microscopy Society of America and the Microbeam Analysis Society, but seems to have vanished. It was developed by Ralph Waldo of GM Research Labs. It runs under DOS. Several people in the EPMA community have been talking about developing a more automated version with a GUI. You can download a copy from my EPMA class website: www. geology. wisc. edu/~johnf/ g 777/777 Software. html

Virtual WDS is an extremely useful program both for teaching as well as for

Virtual WDS is an extremely useful program both for teaching as well as for daily quantitative microanalysis with an electron microprobe. Developed by Andy Buckley and Stephen Reed at Cambridge, it simulates wavelength x-ray spectra-- and all the complications that arise in real life -- overlapping peak and backgrounds, including higher order interferences. Thus the mysterious “pulse height analysis” concept can be more easily visualized by students. This program operates from a database of actual WDS scans from the Cambridge SX 50; they can be also plotted in Å and ke. V units. Academic License is $250

Consider the measurement of 2000 ppm of Mg in apatite with TAP (set up

Consider the measurement of 2000 ppm of Mg in apatite with TAP (set up in the Periodic Table display, left). Green=potential first order problems; Blue=potential higher order problems.

(Left) Virtual WDS immediately alerts the user to the presence of a 3 rd

(Left) Virtual WDS immediately alerts the user to the presence of a 3 rd order Ca Ka peak (red) just to high side, a potentially bad background position.

Pulse Height Operation (Top Left) The PHA settings for this scan, set as integral

Pulse Height Operation (Top Left) The PHA settings for this scan, set as integral (no window), the reason that the 3 rd order Ca counts are very high. 100% of the 3 rd order Ca Ka x-rays are counted. What would setting a PHA window (=differential mode) do? (Bottom Left) Setting a window (3. 2 v), VWDS shows that now only the Ar-escape peak (~6% of the Ca counts) remain, and the plot to the left shows the result: most (not all!) of the higher order elements have been removed.

In differential PHA mode, the 3 rd order Ca Ka x-ray counts are reduced--but

In differential PHA mode, the 3 rd order Ca Ka x-ray counts are reduced--but not eliminated

And this is where he says something about how good this all is and

And this is where he says something about how good this all is and how the students think these things are wonderful and they wish they could spend more hours playing with these programs, that remind them of the video games they played in elementary school …