MC_Demo - Monte Carlo Modeling for WIN98 Introduction The program MC_Demo is a group of simulations illustrating the principles

of the Monte Carlo modeling of electron-solid interactions. The programs were written by David Joy and are based on the algorithms described in the book Monte Carlo Modeling for !lectron Microscopy and Microanalysis published by "#ford $niversity %ress &'(()*. MC_Demo is in the public domain and may be freely copied and distributed. +f you use results from this package in a publication or presentation an acknowledgement would be appreciated. This program is designed to run under ,indow () or (although it should also operate without trouble under ,indows ..'. +f you downloaded the program be sure that the files MC_Demo.e#e and /0122.3.dll are placed in the same subdirectory. To run the program choose 4un from the 5tart menu and use the 0rowse command to reach the directory where the program is stored. 6or e#ample7 if the program is on a floppy disc7 then in the 4un edit bo# enter a89mc_demo.e#e and then click on 4un . The screen will appear as shown in the illustration. The five programs in the demo are accessed by clicking on either the ,alk 7 :!M 7 5!M' 7;5!M3; or %hiro< buttons. +f you want to print any of the displays simply hit the %rint 5creen key while holding down the :=T or :=T>5?+6T keys. The screen is then copied to the clipboard. To view7 edit7 or print these images open the %:+@T program - which is usually supplied free as part of the ,+@D",5 installation in the :ccessories file - go to the !dit menu and then click on %aste. The program obeys all the usual ,+@(A rules. To change any of the parameters7 such as the !nergy7 click the Mouse in the te#t bo#7 then either use the delete key to remove the old value7 or hold down the left mouse button while dragging across the numbers7 and then enter the new value reBuired.

Running the Programs (1 W!"# This program has nothing to do directly with electron interactions but it is a Monte Carlo simulation and it illustrates some of the concepts of using this techniBue. +t seeks to answers the Buestion - how far away from your starting point would you be after taking @ steps each of eBual length in random directions. To run ,:=C firstly choose the number of steps that you wish to model by entering the appropriate value in the @$M0!4 window . Then click on the ,:=C button. you can decide how many steps to take. The program then starts and the display shown here appears on the screen. The display shows the progress of the walker and circles representing '27 32 and .2 steps from the origin are placed on the screen to give an idea of the scale. :t the end of the walk the actual distance from start to finish is computed and displayed in the information bo# immediately underneath the display window. 5uggested e#periments and Buestions8 &a* select a Dourney of '22 steps and run the program as described above. @ote down the reported value of the distance traveled. @ow repeat this operation ten times. ?ow do the values compare E ,hy are they not all the same E &b* @ow repeat these operations for 3227 1227 F227 and '222 steps. +f possible plot the distances for each trial and number of steps on the same graph. Does the distance traveled in a single trial vary in a simple manner with the number of steps E ,hy not E &c* ?ow does the average distance traveled in @ steps &i.e the mean value derived from the ten tries* vary with @ E Gou should be able to show from your data gathered in these e#periments that the average distance traveled in @ steps is eBual to the sBuare root of @. +n order to demonstrate this however

we must perform enough trials at each number of steps to get a statistically valid answer. : single run of the program is not sufficient to provide meaningful results. +f you keep the same number of steps and simply re-run the program M times then the variance of the result will be eBual to the sBuare root of M. 5o if you run the program 1 times with the same conditions the e#pected precision of the average distance calculated is only )2H. +f the same conditions are used for a hundred simulations then the precision of the average distance traveled will improve to '2H7 and so on. ($ !%M 55_MC is a single scattering Monte Carlo electron simulation for thin foils. This is7 within the overall limits of the appro#imations used7 the most accurate model available. The program is designed for a thin sample. 5tart by selecting a beam energy in the !nergy &ke/* bo#. '22ke/ is a good value. To select a sample material use the scroll bo# marked Target Material and click the up or down arrows to reach the specimen of choice. Then click on that material and it will be highlighted as shown above. 5elect a thickness for your sample in the bo# marked Thickness &:* - '222 &angstroms* would be a good choice. 6inally choose how many traDectories you wish to run by editing the contents of the @umber bo#. 4emembering the previous e#periment7 the larger the number that you choose the more precise the result will be. : suitable minimum value would be )22 to '222. ,hen you have set up the values you want click on the :!M button and the program will run with the display shown. The foil is shown as a cross-section with the beam entering from the top and e#iting through the bottom. The thickness of foil that you selected calibrates the scale of this figure. The electron traDectories are color coded to demonstrate any loss of energy that occurs. The information bo# below the display window computes the beam broadening radius - which is the radius within which (2H of the transmitted electrons lie as the e#it from the sample. This effectively defines the e#pected spatial resolution of an A-ray analysis in this sample.

5uggested trial applications8 &a* $sing copper as the material7 observe the beam broadening for an incident energy of '22ke/ and for sample thicknesses of 3)27 )227 I)27 '2227 ')22 and 3222 angstroms using )22 traDectories per run. ?ow does the broadening vary with the foil thicknessE ,hat happens if the beam energy is increased to 322ke/E &b* 4epeat e#periment &a* but for foils of carbon7 silicon7 silver and gold. ,hat is the effect of making the foil of an element of higher or lower atomic number E &c* $sing '222 angstroms of copper as the target observe the traDectories as the incident energy is changed from '22ke/ to .27 ')7 '27 and ) ke/. @otice that thin foil starts to behave like a bulk specimen as each electron is scattered many times E. @ote that at low energies each traDectory may take a considerable time so select only )22 or '222 traDectories and be patient. (& P' %5 is a plural scattering Monte Carlo model which is well suited for studies of electron interactions in bulk specimens &that is7 specimens which are not thin enough to transmit the electron beam* and at low beam energies &J32ke/*. +t is slightly less accurate than the single scattering model &5!M3* but much faster. 5tart by selecting a beam energy7 a material7 an angle of incidence &2 means the beam hits the surface at (2 degrees* and the number of traDectories reBuired. The Thickness value is not used in this module and so this number has no effect. ,hen everything is set up click on the 5!M' button and the program will start to run. The information bo# below the display window will initially show the calculated 0ethe electron range &in microns*. This is a measure of the average distance traveled by an electron within the specimen. The display carries a calibrated micron marker to aid interpretation of the computed interaction volume. The

electron traDectories are color coded to show how the electron loses it energy as it travels. ,hen the chosen number of traDectories has been completed the information window displays the calculated backscattering coefficient. 5uggested trial applications8 &a* $sing 3)2 traDectories and a beam energy of ')ke/ plot the interaction volumes for carbon7 copper7 silver and gold. ?ow do the interaction volumes vary from one element to anotherE &b* $sing at least )22 traDectories compute the backscattering coefficient for copper at energies of '7.7)7'2732 and .2ke/. Does it varyE ?ow do these numbers compare with those for carbon and gold at the same energiesE ( '' 55 is another e#ample of the single scattering algorithm here given in a version designed for bulk samples only. "nce again the relative energy of the beam is shown on an arbitrary color scale. The operation of this program is that same as that for %57 but the routines run more slowly because significantly more calculations are involved. To compare this option with 5!M' re-run the trial applications suggested above.

P)IR*+

This program computes the way in which the yield of a given A-ray line varies with vertical position from the surface. This is called the (<* & phi ro <ee * profile. +f this yield is normali<ed by the eBuivalent yield for the same A-ray from an infinitely thin region at the surface of the specimen then the resultant ratio is called f&* & f chi *. This Buantity is of fundamental importance in the development of methods for the Buantification of A-ray data from solid specimens. To run the program7 select an energy7 material7 and number of traDectories as before. +n addition the KTiltLT":; bo# allows selection of the A-ray take off angle into the detector. +f no value if given then a default value of 12 degrees is assumed. %rovided that the material selected has an A-ray of suitable energy to be e#cited then the program will run. then click on %?+4"M . The program displays a simulation of the production of A-rays using a routine originally devised by Curgenven and Duncumb in their classic paper. The density of dots gives an indication of the local rate of A-ray production but is not necessarily directly proportional to the A-ray yield. @ote that the A-ray production varies considerably both laterally and in depth. ,hen the run is finished the histogram is plotted on the left hand side of the screen starting at the surface. The length of the bars shows the relative of value of (<* at that depth. The bar at the surface is of unit value. This histogram is plotted at vertical increments eBual to '>)2th of the 0ethe range. $se at least )22 traDectories to give reasonable statistics. ,hen the program has finished click the mouse inside the information bo#. The screen will now redisplay the (<* histogram7 on a properly annotated scale7 together with the corresponding f&* histogram. f&* shows what fraction of the generated A-ray production actually escapes from the specimen after allowing for absorption in the specimen using the take-off angle selected during the set-up. @ote that at low voltages the two histograms are very similar7 but when the beam energy is much higher than the e#citation energy the f&* curve falls away because of strong absorption for A-rays coming from deep into the sample. The information bo# reports the numerical value of f&* as well as the volume &in cubic micrometers* from which (2H of the total emitted A-ray signal is generated.

5uggested trial applications8 &a* for copper at 32ke/ obtain the (<* curve . @ow repeat the e#periment for an incident energy of '2ke/. ?ow do the two curves compareE ,hy is there a differenceE &b* Compute the (<* curve for carbon at .7 '27 and 32ke/. Compare the (<* curves at these energies7 as well as the f&* value. ,hy does f&* decrease as the beam energy is increasedE