Hello and welcome. My name is Charles Zona and I would like to thank everyone for attending today’s McCrone Group webinar. Our presenter today is Joe Rebstock of McCrone Associates. Joe is going to tell us what constitutes a knowledgeable scanning electron microscope operator from simply being, what he refers to as a “knob tweaker”. Joe is a Senior Research Scientist with McCrone Associates and has nearly 40 years of experience. He is a co-instructor for our basic scanning electron microscopy course, a course he has been teaching for over 15 years. During that time, Joe has seen his fair share of knob tweakers and has been able to transform them into knowledgeable SEM operators. Joe specializes in particle analysis, polymer imaging, electronics and medical device analysis using SEM, EDS, and WDS. Joe will field questions from the audience immediately following today’s presentation. This webinar is being recorded and will be available on The McCrone Group website under the “Webinars” tab.
And now I will hand the program over to Joe… JOE: Thank you very much, Chuck. Welcome to everybody that is able to join us today. Hopefully, we will have a little bit of information for everybody to take back and find useful in their particular operations. So, the topic today is on scanning electron microscopes, and our question is, “Do you think you are a scanning electron microscopist?” Well let's find out. In today's instruments many functions that are being performed can be done in part or in whole by computer control. This is able because the manufacturers have supplied a number of what they call recipes with the software in the instruments. And these recipes, if you click on them, will set up the instrument for certain conditions, like the column accelerating voltage, the spot size, the current in the beam, the working distance, everything is already preprogrammed for you. So, with the advent of recipes, and these can also be developed in-house, just about anybody can learn to turn on an SEM and acquire information.
The amount of training that anyone of us has received just varies considerably. If we are looking at a work environment where were using the recipes, than we just need enough information to be able to turn it on, acquire the information, know whether not maybe it's running correctly or not. But, then produce results and we’re done. Also, if we have very strict protocols, maybe we have a legal case and everything we analyze of one particular material or type of testing needs to be done under set conditions. Any requirements then for the job will vary greatly from place of business to place of business. So, therefore, if we can operate that SEM according to our protocols and requirements, we are scanning electron microscopists But, if we have to operate the SEM outside of those conditions and parameters, if we don't have enough training and experience, we very quickly become frustrated. We’re trying to develop new procedures, investigate some new areas. And, too often or very often, when we get students what we hear from them is, “If you have to operate under those situations where you start? What do you change first and make it work better for you?” So, here we can end up, anyone of us could end up, as what I call a “trial and error knob tweaker.
” If we have enough experience and have been given enough training, then we become the knowledgeable operator. There's lots of benefits to working at that. So, if you're starting out, what you want to find out is what type of SEM do you have? You want to know the make and model of that SEM. If it is a tabletop you are going to have very limited parameters and controls, if it’s a tabletop SEM. You go to the standard type SEM, which operates in standard vacuum conditions, you’re talk 10-5 and 10-6 torr, you have a lot more controls available to you. If you have what's called variable pressure SEM or low vacuum SEM, this is just an extension of the standard SEM. It allows you to look at samples now that may not be fully conductive, where for the standard SEM you only have one option available. The sample either has to be conductive or you have to find some way to get around the sample charging.
So, you can put nonconductive samples into your SEM with the variable pressure and you can dissipate the excess electrons that are building up in the sample. They also have what is called an environmental SEM. There are not a lot of these out there today. As a matter of fact, over all of the years we have been teaching the course, we never had anybody come take the course that actually had one. So, they have very limited use. It's a slowly developing field. It has some very specific uses, but if you wanted to look at something like tissue of materials, and you wanted to maybe heat it or cook it or something like that, and re-examine it, you can look at it in different stages, but not a lot of demand. The latest SEM's and the one to have the greatest resolutions to them are the field emission SEMs. And, these are now available in both the cold or thermal emitters. So, if you get this information together, what it is going to tell you is you will know some of the limits and capabilities of your SEM, and what you can actually expect to get from it. You also want to find out what the electron gun or electron emitter is in your SEM.
Many of them are just going to be the hairpin tungsten filament. This is just a tungsten wire, it is about 125 µm in diameter. You also have a Lanthanum Hexaboride crystal, which is referred to as LaB6 or LPG. You have your field emission guns and these could be the cold or the thermal variety. This is just a very fine sharp tip tungsten needle. So, just a quick comparison to the type of emitters. This is the hairpin tungsten filament and the field emission guns. Your LaB6 will kind of fall in the middle of these two. But, the tungsten filament, as we mentioned, was about 125 µm in diameter, but the area that could be producing the electron beam that you are going to get off if it is only 20 µm in diameter. That is going to be the diameter of the cloud or the beam of electrons that is being emitted.
If you go to the field emission gun, it is considerably smaller. Now we’re talking about area on the tip that is less than 5 nm in size. So, with the smaller diameter beam at the gun, we can keep it finely focused and when we get to the sample we can get fantastic resolution out of it because it's super small. So, the magnifications available often on the standard type SEMs, the LVs, is up to about maybe 300,000X and achievable at about nanometer resolution. This is only if you have everything perfectly aligned and you have the world's best prepared sample. So, you might really anticipate getting something more on the order of between 25,000X and 75,000X in most cases. If you go to the field emission, they have the capability of going to 1,000,000X. This is takes a lot of effort to get to. You have to have a room that is well protected, no emissions of electromagnetic emissions in the field anywhere near the instrument, otherwise it will distort your image.
But you can achieve resolutions of less than 1/10 of a nanometer, you’re talking now that you are down into the Angstrom range. So, other things you should know — you want find out what detectors are on your SEM and where they are located. Where they are located is going to play a big factor in interpreting your data. So, do you have a secondary electron detector, and if so how many and where are they? Do you have backscattered electron imaging capability? Typically this is going to be what is called the composition mode, designated maybe as BEC. But if you have another detector right adjacent to it, you can add a condition to it that is called shadowing. That is BES designation. And it actually adds quite a few features to the images, it’s very very good. And there's also the PET which is looking at topography. These detectors interact with one another, signals are added and subtracted to get the image, but it gets you a lot of useful information.
You also want find out if you have an energy system on it, so you can detect x-rays or wavelengths. You could have both. Where are they positioned? So, after we turn on our SEM, we have it under vacuum and it’s ready to go. The first thing you want to do is saturate the filament. Well, some people have said, “We don't saturate filaments. It's just on when we hit the button.” Well, what we recommend is that you go into your gun and you actually turn the current that is in your filament down when you go to enter your session. When you bring it back up, start at zero and you're not shocking this filament. It's very delicate. It also forces you to recheck your saturation point because as a filament is used they wear, so, therefore your saturation point may change over time very quickly.
So, it won’t be the same. You may think it is, but your image is going to degrade on you just by turning it on and off You want to find what is called the heating current, or load current, or filament current setting. The terms are synonymous. The same manufacturer will switch between instruments. They may use the abbreviation for it or maybe even acronym, like load current might be designated somewhere your software as LC. So, to saturate we use our little diagram here, our graph. The amount of current that is going into the filament circuit is set represented by our X-scale. So, as we put current into that it heats the filament. Initially it is being heated very close to the post where the tungsten wire is attached. As we turn that current up more heat is generated. It will eventually maybe get up from point A to point B.
And, what we’re looking for in our monitor is the image starting to become brighter and more intense. So, you’re seeing brightness and contrast showing up. After point B, we heat it a little bit more and the image starts to get darker. I've had people actually say they stop at position B. You don't want to operate here because the beam is not focused. You're getting a false image here, a false saturation, because the electrons are being emitted from the two sides of your filament. They need to come off the tip. In order to get to the hottest, you have to turn the current up even higher. So, now you go through position C, D, and E; and you get up to F and all of a sudden, now your image is maximum brightness. It's not changing at all. This is now saturated. You may be oversaturated, and you don't want to be here either.
In order find out your oversaturated, start dropping the current and look for the least change in brightness and intensity. This is what's called a shoulder to the plateau. If you run your instrument on the shoulder, you can extend the life your filament by many many hours. So it will operate longer for you. Now, if you need better resolution later on, you may have to come back here and go back to saturation. After we’re saturated, we need to align the beam as far as how it's passing from the emitter down to your specimen. It’s called column alignment. One of the students in our class came up with the acronym of GAS. They said the first step is to align the Gun. Go all the way back to the filament. And, what you want to do is to adjust your shift and tilt controls. It should be designated somewhere in your system. As such that the beam is coming off the very tip. You want it to be on center, not off to a side or in either direction.
You also want to come straight off and go straight down the column, that's the tilt. You do not want striking the wall somewhere along the line and getting a black image. You need to have it point right down the column, go through all the openings all the way down. The next adjustment is what's called the aperture. We will say a little bit more about that in a second. But, for the aperture, we want to be at best focus. You want to find a small feature, and you want to blow it up to higher magnification than when you were at the adjustment for the gun. For the gun, I like to use low vac. So now I have an image maybe an inch or two in size in the center of my screen, I am at best focus, now I look for a control that is called wobbler.
And, what it does, it’s an electronic circuit that is going to focus and defocused above and below the sample electronically. What I want to look for in my image, the feature that I am observing, is issued symmetrically — going in and out of focus about itself. It should not move on the screen, you know, left, right, up, down, or anything in between. It needs to go in and out of focus symmetrically. The best type of feature to use for this is actually going to be a sphere. It works great. So, now we adjust the aperture, center it. And some people may or may not have this, we will get more to this in a second. But, you center it, and then turn it off. The wobbler will not stop at best focus. You will have to readjust your best focus, because the next step is adjust stigmation. What the stigmation is going to be is the shape of the beam.
The cloud of electrons going down column —when we get to the sample, we want it to be circular in shape. If the beam is not circular, if it's elongated in one direction or the other, it’s going to be elliptical. And what will happen is there will be preferential stretching in our images. So, we need to remove that, and it should have a set of controls somewhere on your keyboard or in your software, that are XY stigmator controls. You bring both of those into the best focus again. And, remember as you use your SEM overtime, continue to check best focus and continue to recheck the stigmation of your image. The SEM is somewhat unstable, especially the first 30 minutes of operation, and these will change frequently on you. So, continue to recheck them. You set them one time and they’re eventually going to become very blurry.
So, now what you've done your column alignment, there's a couple quick tests you can do. Here's the size of the feature I usually use and about the approximate magnification for the first cut alignment. I am about an inch or two in the center of the field of view of my particle and about 5,000X. If you take your spot size, which is your beam current, and it changes the diameter of your cloud as you turn it up or down. So if you change a couple clicks over where your current setting is, and the go below it, what you want to observe is the image and how it remains positioned on screen. Does it move left, right, up, down? If it does how much does it move? Typically, I am doing my alignments or final alignments 3 to 5 times the magnification I think I want use to acquire images. Now, if I’m three times greater than where I am going to get my image, if I have some movement of a millimeter or two, I don’t worry about it, because when I back off, and go to magnification I plan on using that little bit of distortion is not seen. So, if it does move, the adjustment is in your gun.
Go back to the gun, recheck your saturation, recheck your shift and tilt controls. The other control, or test rather, is going to be the change of focus — a few clicks above best focus, a few clicks below best focus. And, again, watch your image. If it moves, then you need to make another adjustment. This adjustment is going to be back at your aperture. Of course, anywhere along the way keep your eye on stigmation. Stigmation is just stretching of the sample in a preferential direction. Let’s look at our particle, here. We did a saturation. And, the image on the left represents the particle before the gun alignment. Where on the right, now, we have the alignment completed and we now can see a lot of structure and detail. Just very quickly, here is a cut away of a column. The gun is near the top, our condenser coils are in center here, in the green area. And just below it is the location of your aperture.
The SEM's we use this would look like a representation of that aperture. For some people it may just be a box. You may not even have this control available to you. Either it's not on your SEM or it is being corrected for electronically, for most of ours we use a physical adjustment, but we do have some that are electronic. So, here is our aperture and on this particular unit there's multiple numbers here near the top right, you see 0,1,2,3. These are the positions that you can put this aperture in. What’s gonna happen is when you move from one the other, all the way on the left in the circled area, it's where there is the end of the aperture that is in the electron beam path. Go down below that and we have a little flat piece of metal represented here. And, it’s pointing at it and it has the 100 µm symbol below it. This is a piece of the molybdenum (Mo) metal, and this is the actual working part of the aperture that is going to get our beam where we want it. There’s holes cut in it that are shown by the dots that are represented there. In the zero position on this aperture, it is totally removed, there is no aperture in the beam path at all. In the number one position, the whole that is in the beam path is 20 µm in diameter and you want to use a small diameter aperture when you're trying to acquire high-resolution images.
The largest position is a number three aperture, and it is 100 µm. Here we want to get a lot of electrons down a column all at once and we want to try to collect elemental data. So, we are forming the EDS spectrum or collecting WDS counts. The position in between, the number two position, is 30 µm and this is one the manufacturers put in and it is a kind of nice intermediate between the two. So, you can do some nice imaging with a number two setting, as well as do some good elemental work. Don’t think that the number one position is always going to be the smallest hole. I've seen in the same manufacturer be the opposite way. In one instrument the number one is the smallest, and in a different instrument by the manufacturer number one is the biggest. So, check it out. This I have thrown in just as a quick example to make you aware of a little problem that you could come up with. It’s inherent in the instruments.
This is the aperture, the molybdenum metal piece, and there are three analysis done here near to this one hole. Of interest then, is the amount of carbon that is turning up here. All three positions turn up a fair amount of carbon, and the heaviest areas closest to the hole. Well, what big deal is it if there a little bit of carbon on the surface? None, it doesn't make any difference. But, this carbon is due to oil vapors that are in the vacuum and these are coming back from your pump. You’re talking 10-5 to 10-6 torr here of vacuum. So, you are able to pull some of the vapors out the pump back into the instrument. Those vapors get back into the column and they interact with beam and the get burned on all of these surfaces. You can run into a problem when this deposit starts to build up at the periphery of your hole.
It does not build up symmetrically. It'll become irregular shaped in a big hurry and it will really be blocking your beam. You are going to suffer a lot of degradation. This aperture does need to be replaced periodically. So, if you don't have a PM routine from the manufacture someone at your facility should be routinely looking into this to change it. As the beam then interacts with the sample it generates multiple types of radiation signals. The ones that we are a go to be interested in for the work that we’re going to do on the SEM, some on the left, we have couple here, is called Bremsstrahlung radiation. We will see more in a second about these. And then, our characteristic x-rays, we have our secondary electrons shown on the right, along with backscattered electrons. But, you cannot overlook the fact that you are going to have heat. Anytime the beam interacts with the sample, you’re passing electrons from one position point to another. So, when you get electron flow, you always have a current and it’s going to generate heat. It is going to be possible to burn your sample. You could maybe melt it, cause it to change form.
You could even vaporize some sensitive samples. So, consider your heat, and a way to get rid of some heat is to lower your current. Take a look now at where our usable signals are coming from. As our beam comes down to strike our surface, it is the purple cone right up at the top of the image, and it rasters over that small area at the bottom of the cone. The electrons do not stop at the surface of your sample. They actually penetrate into it; they’re impinging into the material being analyzed. And, then once they're inside, depending upon the material you're looking at, they begin to swell, and they might swell immediately or not quite so quickly, as this example is showing. And within the area then of the excitation, you have all your signals of interest being generated.
They are generated throughout the excitation volume, but secondary electrons have very low energy. And the only ones that are really useful to us, that we can get out of our sample, and back to our detector are those that are very near the surface service. This may be on the order of maybe just 10 to 50 nm into the sample. So, we end up with very nice image quality here. Potentially, we could get very good image quality and a lot of surface structure detail. The best area, or the most usable area, to get your backscattered electrons — the bulk of them are coming from this green area, and they are of higher energy than the secondary electrons, so therefore they can get out from a deeper depth. We have some examples for this. And, then below that, the bulk of our x-rays are generated in the purple area near the bottom. These signals are generated throughout the excitation volume, but the bulk of what's coming out that we can get to our detectors, that are usable to us, are in these designated areas. So, you have to consider this excitation volume and shape, because if you were to try to analyze, say a laminate, and you're looking at in a cross-section, you may be rastering on one of the layers on that laminate but because of the subsurface swelling in the volume, you could be getting a signals coming from adjacent layers.
So, you have to understand the interaction between beam, the specimen, and your detector. How do they form their signals? We have an electron coming in from the beam and anytime an electron’s path is changed it'll slow slightly, but it gives off a little bit of energy. And if we have a very small change, we get a low energy x-ray. And, these are not characteristic or diagnostic, the change is random, and therefore this what forms the background in our EDS spectrum. If it bends just slightly these are referred to as inelastic electrons. Where if it is caused to turn at a much higher angle then these are elastically scattered electrons. And our backscattered electrons are actually elastically backscattered electrons. Its beam electrons go into the sample and are being ejected straight back out. Over on the right side in our diagram here, we have an electron from the beam coming in and it strikes an electron that is orbiting one of our atoms.
And, it actually dislodges it, and the dislodged electron from the specimen is now our secondary electron. We can collect these if there's ejected from an atom that’s near the surface and we get a very good image generated from it. The electron from the beam is inelastically scattered and it goes off and it interacts with other parts of the specimen or it gets absorbed. Now once we knocked an electron out of orbit, that atom is unstable and it will stabilize itself in short order. What will happen is an electron that is orbiting in another shell will be pulled from an outer shell to fill our void. And, it cascades and continues, but electrons from outer orbits must have higher energy to maintain their orbits than the inner electrons. So, therefore what happens to that energy? It’s shed in the form of an x-ray and the amount of energy that’s in that x-ray now is a diagnostic, and characteristic information.
It will tell us what shell it came from and what element it came from. Just a quick example here of EDS spectrum, we’ll come back to this later. But, if you look at the area, let's say in between about 2.5 where the molybdenum peak is, and then just over 5 where a chromium peak is, you can see what the backgrounds in our spectrum is going to look like. And these will taper off to either side. But, anything below our peaks is background, and it actually gets subtracted out in the calculations. The peaks then are representative of the elements that are in our sample that we’re analyzing, and the peak intensities are dependent upon the concentration for each of those elements. You can have multiple peaks for chromium, manganese, iron, and nickel. These are because these x-rays are coming from different shells or subshells. So, for the two nickel peaks we’re looking at here, right around 7, a little over 7.
5 and then little over 8.5, or just short of 8.5, the first nickel peak is called your K-alpha and your second peak is called K-beta. Those x-rays were generated in the K-shell and the electron that filled the void came from one orbital away to form the K-alpha, and it came from two orbitals away to form the K-beta. And, there's also nickel L-x-ray lines down here at the far left. So, there’s a number of the x-rays that can be generated from different energies from each element. Anything over about silicon. So, now if we have an image acquired, we have gone through and done an alignment, it’s fully saturated, the sample is well-prepared, but we’re unsatisfied with the image that we’re seeing what can we do as operators to improve that image quality? Well, one of things you can do is change your accelerating voltage. As you increase the accelerating voltage, the diameter of your beam actually becomes smaller: Well, when it becomes smaller, the result is you get better resolution, but don't get into thinking that better resolution always equates to better image quality.
We will see just the opposite. We could change the spot size, which means the beam current. As we turn beam current higher, now beam diameter gets bigger. Well, if we turn it down for imaging purposes, the diameter will become smaller. That will give us better resolution, and we can also use a better image quality out of that. We can change the working distance and this is the space in between the end of the column and the focal plane of the specimen where were currently viewing. It's measured in millimeters. We can also change the tilt, the way that the specimen is configured to interact with the beam. So, if you can tilt it just a few degrees, sometimes you can get an amazing effect as far as image quality. We have some examples here. Also, you want to take a look at our objective lens, if this is available on your instrument for adjusting. As you change the size, you can affect the image quality.
We will show some examples here. First, I want to take a look at what happens to our excitation volume. If were looking at a low atomic number material, the typical shape is — let’s say we’re looking at a piece of carbon— we get what is called a teardrop or pear-shaped excitation volume. We’re rastering the area at the surface between the two arrows, but then it impinges into it, goes down a little ways, and then starts interacting with the electrons and protons in the atoms of the specimen, and it swells. Well, if we go from low accelerating voltage, only at the top, to a higher accelerating voltage, maybe 5kV to 30kV, the excitation volume merely gets bigger. It swells, it gets deeper and wider. So, always keep that in mind. If we take a look at a higher atomic number material, maybe a metal, basically what happens is that the upper half of the excitation line we were just looking at in the carbon sample, basically goes away. It gets cut off, sliced away, because there's so many protons and so many orbiting electrons, as soon as the electrons from the beam strike the surface, there is almost instantaneous interaction and scattering. So, the excitation area again is in the middle of the surface. It is in between the two arrows.
But, were getting x-rays generated from a pretty good distance away. An example of the penetration of the beam is shown right here in a TEM grid. The TEM grid on the left, this image was acquired at 1kV. This has an organic film cast on it, and it’s just a TEM grid. If nobody has seen it, it's just an array of squares, open squares, and the carbon films or organic films are placed there so that you put small particles on the film and not have it fall through. Something to keep it there. So, then what you're doing is looking at specimens in the openings between the grid bars. On the right, what we have done with this image is we have increased the accelerating voltage, and this is now up to 5kV. And, the beam is actually penetrating now through that organic film. It was quite thin, sub-micrometer in thickness. So, initially it was being stopped, and now at 5kV we are going through it. We can see through the grid squares and we can see small particles on the surface of the substrate in between the openings. Also, here is a paint chip and penetration effective beam on this paint chip.
On the left is 2.5 kV and we get very nice structural detail of the surface. We can we see small features on that surface. But, if we increase the accelerating voltage to 20 kV, the quality of our image has degraded. Well, at higher kV, we have a smaller diameter beam, it should give us better resolution. Well, resolution is not image quality. What is happening is we’re impinging too deeply, we’re getting too many signals from too many planes and we lose the structural of the surface. So, if you are looking for imaging, you’re going to back off, and go to that lower kV. If you’re going to do elemental analysis, maybe you need this higher kV. So, the optimum conditions here for imaging are not the same as our optimum conditions for our doing elemental work.
So, we’re going to want to try maybe another accelerating voltage and see what happens. But first, let's take a look at the same secondary image on the left, at 2.5 kV. And, we switched it from the secondary detector to the backscatter detector. Well we don't hardly see a particle. You can barely make out there's something there. There's just not enough signal under the other conditions. We could change something to improve on that. Change maybe of brightness/contrast, or maybe change the amount, the number of electrons going into the sample without changing kV, just to create more signal and start to bring this particle out. But, another way to do it is just change our accelerating voltage. Well, we will we've gone from left from 2.5 up to 10kV. We can see some very nice structure in the backscattered image. And this is the compositional mode so, we can see some different brightness/contrast levels showing up.
And, if you are doing elemental analysis on these, you’re going to want to examine the different brightness/contrast levels and make sure exactly what's going on. Because in your backscattered image, the compositional mode, the actual brightness/contrast is determined by two factors. One, the composition in any given point and its average of all of the elements that are in that point. Or, the topography or the surface structure of the sample. So you could actually have black holes in the surface of the sample and there is just no material there. But, nevertheless, it's got a different brightness/contrast level. If we go over to the right on the same paint chip, we boosted the accelerating voltage now the 20 kV, and we have these bright features showing up. In the little particular paint chip we have here, we now have some lead particles that are associated with our paint chip. They're not on the surface because we could not see the with the lower kVs. So, therefore they are subsurface or they may be embedded within or at the opposite side of our paint chip. We can turn it over and find out by looking at it and see we get a better a secondary image of it.
In the secondary image there was no indication they were even present. So, you need some different conditions. Use your different detectors. Here, we want to take a look at what affects your depth of field. We have a sample on the left and right, both of these images are at 25 kV. We will keep the kV the same. We are looking at spring. Right now the top of the spring is 10 mm working distance away from the column and the spring is 4 mm thickness, as far as top to bottom, side to side. It is being viewed with a aperture of 100 µm, so, the largest opening aperture. How can we improve this image so that we can see the top of the spring and the bottom of the spring at the same time? We want to increase that depth of field. One of the ways is to change the working distance. We actually lower the specimen further away. Here, we have gone 10 millimeters to 50, so that’s a change of 40 mm further away. That is the only condition or parameter we have changed, and look at the difference in image quality were getting out this.
We can actually see small particles now on the bottom of the spring. In this example, we are back at our 10 mm with our spring. So it's distorted just like it was in the first example, on the left. But, here, we change our aperture. We go from the largest opening, the 100 micrometers opening, to the smallest opening. And, again we get a significant improvement in the depth of the field. We can again see those fine particles on the bottom of the spring. One more time with our example on the left, back at the starting conditions and our optimized conditions now on the right. We have changed both the working distance and the aperture and have greatly improve our image quality. So, in the SEM we have the option of tilting the sample to improve it. What's happening is on the far left we have our surface of our sample, normal to our beam and the beam is impinging into and forming a normal expectation volume. But, if we take our sample and tilt it or tilt our stage, if you have this capability, tilt the stage. Then what happens is the excitation volume is brought closer to the surface and because it's brought closer to the surface, now the region generating the secondary electrons, we get more secondary electrons being emitted.
So, they have less solid material to pass through. We have artificially pulled this excitation volume closer. And just a few degrees of tilt can make an amazing difference in your image quality, if you can use the tilt factor. If you don't have a tilting stage, just take your substrate and configure some way to put something on one side of it and tilt over a little bit. We also frequently get asked, “How do I tell the difference between a protrusion in a field of view and a depression or a hole?” Well, if we have a protrusion the side of the feature sticking up that has a direct line of site to your be is going to appear brighter than the opposite side of that same feature, just as shown here in this example. We get the direct line of sight on the right, because our detector, in this example, is sitting up at about 11 o'clock. The side of the feature protrusion, from the right side of it, the electrons coming off must go up and then over the top of the protrusion.
Some of them just aren't going to make it. There are going to be many, many fewer secondary electrons escaping and getting to the detector. So, this is your example for a protrusion. If this was a depression, a side or edge of the feature closest to your detector, this time, is going to be your darkest side. Because along that hole or that wall close to the detector, those secondary electrons have to come up it over the edge. You won’t get as many of them. The far side, now, of your hole has a direct line of sight back to your detector and that edge of the feature will now be your brightest, so the two are the opposite. So, now if you know where your detector is, you'll always know whether or not you're looking at a protrusion or depression. One other feature that can show up in your secondary electron images is called edge effect.
Here you have a surface in your sample or your specimen, grain of sand or whatever it is. Along the edge, it's tilted. It's not normal to the beam. So, it's not perpendicular. It's somewhere in between the flat and vertical. And what has happened is the edge acts as if you have tilted your sample. Your excitation volume that's closest to that surface has been drawn up because of the shape of the feature itself. So, now you have a region on your feature where the use of secondary electrons, of the numbers, have greatly increased, so it appears brighter. Take a look at the example, high versus low k beam. In the square box on the left, this a 30kV image and everything is excessively bright here, all of the edges are really lit up, some of the features in the particle, in the center right there, are not particularly crisp or sharp. They are there, you can make the out, yes.
Inside the box, at the right edge of that particle that's in the square – it actually fades away and looks like it is gone. Well, if we lower the accelerating voltage, so we’re not impinging as deeply, we can decrease the edge effect. You can also turn your current down. That would be helpful. But, now they particle in the square, you can see there is material there at the right side. It's now visible, you can see structure on the surface, the excitation volume is back closer to the surface. So, we see and get more detail. In the darker areas in that particle, those surfaces are more normal to the beam, and the brighter areas are the slanted sides in this particle. So, we are seeing edge effect along that area. Everything though more distinct in this image at 5kV, then it was at 30kV.
Take a look real quickly at microanalysis of analyzing x-rays. Our goal is to use our x-rays generated by the interaction of the electron beam with the atoms of our sample. And we want to identify the elements. If were only identifying the elements present, this is referred to as a qualitative analysis. But, if we go a step beyond that and try to calculate the weight percents for each of those elements then that becomes the quantitative analysis. So, the x-rays are produced simultaneously now along with your secondary backscatter electrons. You don't have to do anything particular other than have the right conditions and start the acquisition. X-rays can then be detected by energy dispersive x-ray spectrometry, and typically in the range of 0 to 20 keV if you use the right accelerating voltages. It is also detectable by wavelength. So, now what you're doing is as the x-rays are generated and they move from the point when the generated to your detectors, they don't just go in a straight line. It is a sine wave type of pattern to get there.
So, with the WDS you are taking a spectrometer and you are optimizing it to look at one x-ray line from one element at a time. It's a slow process. But, whether you do EDS or WDS, both can provide you with qualitative and quantitative data. Your analysis methods then can include using of x-ray dot maps or x-ray line scans over two defined points, start to stop. You could do quantitative or qualitative spectral analysis. You can raster your beam over a set area or one that you define, whether it be a circle, a square, rectangle or one that you randomly draw. Or you can set the analysis to be in a spot mode. Your EDS system can detect all the x-ray lines from the elements that we can analyze from 0 the 20 keV simultaneously, 0 to 20 keV. The WDS is only going to be able to be set up and do one x-ray line per one element individually, it's a long process. Hopefully you won't have too many elements if you’re trying to use this one.
But, the problem with the EDS is it may have too many overlaps for you, How do you resolve them? There are some ways to get around it. One is use the WDS, it very quickly resolves the overlaps in the EDS system, as well as, having a better of detection limit at least order of magnitude for just about all elements. Things to consider —your analytical conditions. What is going to be the beam energy that you're going to use? If you use a minimum of 15 kV, you will be able to detect at least one x-ray line for all the elements in the EDS spectrum that the instrument can detect between 0 and 10keV. If you set your beam current high enough, typically one such that what is called “dead time” is at least 20% or higher. If you have a count rate coming in this below this, your peak to background ratio is going to be very minimal.
Your statistics on your analysis is going to be much lower. So, you want to get as many counts as you can in as short a period of time as possible. But what's dead time? Each time an x-ray strikes the detector, it actually turns itself off, just very briefly, to measure the energy of that x-ray, and then stores that x-ray in the proper bin in the EDS spectrum, and turns itself back on. Well, it is not uncommon you're count rates of maybe anywhere from 2000 to 7000 counts per second or higher, maybe even 50,000 counts per second, sometimes. In order to generate an x-ray you have to have what's called an overvoltage. The overvoltage is where you will set the accelerating voltage, and it must be a factor of at least 1.5 times greater than the energy of the x-ray you're trying to generate. So, if you are trying to look at a copper K-alpha energy line, you're talking about 8keV for the copper x-ray. You would have to have a beam accelerating voltage of a minimum of 12kV.
So. 1.5 times. Where possible. I like to use a factor 2, instead of just 1.5. When you're starting to get near the end of the limit of what you can detect, you can get a lot more counts in less time very easily by having that additional boost in your accelerating voltage You need to consider your sample. Is it homogeneous? If it's not you may have to have multiple analyses from multiple points over maybe a wide area. You have to evaluate each one and set it up to do that. Your peak overlap — these can be difficult to resolve. But if you maybe get a little bit training in this area, and this is the EDS spectrum or x-ray analysis, this is a is a big topic and it will take some work to really be to do it all well and get it right every time. But it's possible But the peak overlaps will certainly have to be considered all the time. It is possible to have contamination within your sample. Or you may get scattering of electrons in your chamber due to nonconductive samples, and they may strike the surface outside your analysis area, and you may be generating signals from the SEM itself. Maybe the mount that you put it on, if you use an aluminum pin, you may be getting aluminum in the spectrum where there should be no aluminum.
Or you might be getting iron from steel chamber. All these things would have to be considered and evaluated. Your software has some options available to you. One of which is going to be the correction method that you're going used to convert your qualitative analysis to your quantitative analysis. That is one of the routines that is often available and you may have the option to choose different ones. One of them is called ZAF and the Z stands for correction of the atomic number, which element is being present in your sample. The A is the absorption, and how many of the x-rays are being absorbed and lost, or fluorescence. You're getting secondary fluorescence coming out of this, maybe from another element, you don't know. But, you also have the option in your software that you might choose to use the manufacturer references to convert your data from qualitative and quantitative. Or you may want to set up and define your own standards.
It is a complicated process, and probably not something the beginner wants to take on right away. So, here's our peak, or our spectrum rather, and this is just an example of a titanium, but one of the things in the EDS spectrum that you need to always consider is what are called artifacts. In this example, it is a piece of titanium, and over towards the right of spectrum there's label there that is called Ti Sum. So, that is a titanium sum peak. It doesn't show up very well in this example. So is it there? Is it not there? Is it misidentified? And the auto-identify routines have been known to misidentify a few peaks. But you as the operator, then, have to verify each of these. If we take that spectrum and convert it from a linear scale to a log scale, we can indeed see there is a titanium sum peak. Well, what is a titanium sum peak? We have our main titanium K-alpha peak at 4.5 keV.
What has happened is two titanium K-alpha x-rays have struck the detector simultaneously, and the detector does not know that it's two x-rays. It sees the total combined energy. So, now it is not 4.5. It is one x-ray energy at 9, so it's 4.5×2. If we had, well we have a small peak here on silicon, if that silicon had a much higher concentration, it would be a much bigger peak. But, if it was a much bigger peak, any of your major peaks then, are really subject to giving you sum peaks, So, a sum peak might be like two titanium x-rays striking the detector simultaneously, and if that was a big silicon peak, you might get two silicon x-rays striking simultaneously. Or you may get one silicon and one titanium x-ray striking it simultaneously, and now you have a different sum peak. So, you could get three different sum peak out of that if you just had two elements. The escape peak for titanium is right here, in between two and three. What an escape peak is, is the titanium K-alpha x-ray has struck the detector, and this is a silicon detector, and in doing so, it creates an electron pair, and they will migrate through the crystal and then on to be measured and detected.
Well, what can happen is these electron pairs, one of the two electrons may become close to an edge and actually escape the crystal, and then it’s lost. The result is you have the x-ray that struck the detector at 4.5, loses the equivalent of a silicon x-ray, which is 1.7, and now you get escape peaks 1.7 keV below your main peak, or in this case 2.8 keV. And you can get rid of some these artifacts maybe by decreasing your count rate. Turn the amount current of down, if you have that luxury. Drop your accelerating voltage if that's a possibility. But, you can adjust for it, and if you can’t then just make sure you account for them properly. If you try to identify some of these peaks, if your software does not label them, and not all manufacturers label every sum or escape peak, you may spend or waste a lot of time trying to identify some strange peak. Here we have a quantitative result.
This is done on a steel standard, and again in this spectrum, we have molybdenum identified over here just above 2keV, and then there's a couple more labels for molybdenum out here between 15 and 20. But, we don't see a peak. Is there something really there? The reason this might come up as a question is because the peak that we see for molybdenum back at just under 2.5 there, this is an area that has a lot of interference. Lead M-line forms an x-ray right there, as well as, the sulfur K-line. This is the molybdenum L-line. So, there's three elements piled on top of one another possibly in that same region. Well, is Molly present or not? Flip to the log scale. We see out here at 17.4 keV, we do have a peak for molybdenum. It is in the right place. And, we look at our spectral result, we see that our molybdenum result, in the second column, has been created or generated based upon the peak in the L-series. So, that's the one back here at about 2.
5. You got your most counts there, you will get your better statistics from there. So you have verified that the peaks are being identified correctly. What you will also want to look for is whether not all peaks have been identified. If we look right here, just about 8keV, right there on the left shoulder of that the nickel K-beta line, it looks like it has some type of a shoulder to it. Is it a doublet, or what is going on? Well, actually 8keV, the element that comes in right here, is the copper, and it would be the copper K-alpha line. In this particular steel actually does have a small amount of copper added to it. Well, in this particular case, it has not been identified. You would have to test it out and find out is that truly copper, or maybe count longer, or do something additional. Maybe use lower kV, expand the scale and see if you get the copper L-line being formed out here. This area gets quite busy and it may not really be distinguishable.
But, it might be another way to check. In conclusion the short course on the SEM operations can provide some significant returns. The SEM operator will be able to make informed, knowledgeable decisions about how to set up and adjust your SEM to obtain better information in less time. You also will have the operator now that is no longer frustrated. They will be much happier worker. A couple of important items to remember. The optimum conditions for imaging are not necessarily the same as optimum conditions for elemental analysis. Often the two can be contradictory. We need high kV for one, low kV for another. High spot size for one, low spot size for another. And, also changing any one setting to improve a result in one area, will certainly change a result, or degrade the signal in another area. I want to thank JEOL USA and Oxford instruments for giving us permission to use some of their diagrams in our training sessions here. They have been very helpful and supportive to us over the years. CHUCK: We have just this slide here that shows some of our upcoming courses and, of course, the future webinars.
And I would like to thank Joe today for a very informative session, and it looks like we have some questions. If you have some questions that you would like to ask Joe at this time, go ahead and type them into the question field and we will the will start answering them. So, the first couple questions looks like it pertains to the availability of this presentation in the future, and this webinar is being recorded and will be available, the entire recording on our website, under the webinars tab, along with the transcript, a PDF of the entire presentation. So those will appear in about a week’s time on our website. Brandon has a question, “Could you explain a little more about the reasons for dead time?” JOE: Once an x-ray strikes the detector and it measured the energy of it, the electronics, in this computer, has is stop and store that x-ray, and just flat turns itself off and it cannot receive incoming x-rays. It's an extremely short period. You're talking micro seconds, here.
But, the software then is measuring it, determining the energy, and putting it in the proper place in your EDS spectrum. Like I say, this can be up to thousand times a second or 50,000 times second it is very fast. But, what happens is you create this dead time, you set up a condition where you want your analysis to run for 100 seconds of live time. Well, live time is only the time that the detector is actually available to measure x-rays, so the total analysis timer clock time becomes a combination of live time plus the dead time. So, now if we have let's say 25% dead time set up for an analysis and because of the number of counts coming in and we count, we tell it to count for 100 seconds of live time. The actual time is going to take is to be the hundred seconds +25 seconds of dead time or on your watch it is going to be 125 seconds, or 2 min.
and 5 seconds for the whole analysis, start to finish. Does that help, Brandon? CHUCK: And kind of related to that, Joe, we have another question from Joe, “Is a rule of thumb or a high limit for dead time?” JOE: Yes. If you have the old-style silicon detectors, the silica lithium drifted detectors, it is possible to harm these things and destroy them because your count rate gets too high. The manufacturers use to show the display of the dead time by color. If it was green you were in the safe zone, if it turned yellow, your count rate was becoming too high, and if it got red you were subject to damaging your crystal. What you are creating, you have got again an electron signal going on here. So, you’ve got heat being generated. In the old detectors, the sili-type detectors, you could actually make that crystal becomes so hot, even though it's being liquid nitrogen cooled, it would become so hot that you could drive the lithium out of your detector and now your $10,000 detector — it just became worth $0.
In the STD you're going to have other complications. You can drive it up to 50%-60%-70%, if you choose. But, if you get count rates this high, what you have to be thinking of then, is the EDS artifacts. Some of the manufacturers try to compensate for this by what they call pulse pileup routines, and they try to subtract it out. In some areas it works fine. But, maybe in the region for other elements, another x-rays and it doesn't work so good and you come up with peaks that you just cannot identify. CHUCK: Ok. Ed is asking, “Is it common to find high carbon on a stainless steel sample using EDS?” JOE: Yes. You can find it. Even at 30kV, you probably end up with carbon in your EDS spectrum. It will be a very small peak because most your excitation volume is subsurface. But, if you drop the accelerating voltage, if you want to know if there is carbon on the surface, drop your accelerating voltage. Bring it down.
If you started the first analysis at 30, do an analysis at 20. If the carbon is from a residue on the surface and, we will get back to that in a second, but if it's from the surface, you should get a higher quantitative result for carbon, because more the excitation volume was closer at 20kV than it was at 30kV. What you're doing is a poor man's depth profile by dropping the kV. If you continue to drop it and go to 10 kV, the carbon result could become even larger. I lost my train of thought on the rest of that. CHUCK: There is one here, Joe, regarding sample size, “How fine of a sample do you need to analyze in the SEM?” JOE: Well, as far as the width and length of it, it can be almost any area, You can limit your raster area down to something that is submicrometer, but depending upon accelerating voltage you are trying used now, if you're looking at a quarter micron particle, and I have in some cases, do not expect to get good quantitative results. Because the bulk of your excitation volume is not being contained within that quarter micrometer particle.
You can identify maybe the major and the minor elements that are present and get a rough approximation of the relationship between them. But, it's not going to be good for all types of analysis. The particles, like I say, I've looked at particles and gotten some hints of what’s present in them below a tenth of a micrometer, but it's not accurate information. So don't treat it as such. CHUCK: If you have a constant accelerating voltage, does changing the spot size have an effect on the interaction volume? JOE: Okay, the interaction volume in size will stay the same. What is going to happen is when you change the spot size, if you increase it, yes, it becomes a little bit larger. But, you are going to lose a little bit of image quality.
You may or may not notice it, depending upon your magnification. But, you're putting more electrons into the surface of the sample, impinging into it, and what you're going to do is create more x-rays in any given second during a time period. So, you lose a little image quality, but not much. You gain a lot of x-ray counts and you get much, much better statistics. CHUCK: Question from Jason, “Any tips for imaging objects that experience motion when excited by the electron beam?” JOE: Coat it, that's one thing. One of the things to be cautious of, if you are looking at a specimen and in your field view, and it is very slowly look like it's drifting out of the center and going to one of the edges, you need to realize this sample itself is not moving — this sample here is stationary. What you're actually seeing is the sample is charging, it is not fully conductive and because it is building up a negative charge, it is repelling the beam as it's coming in.
So, it's being deflected away from the initial area that you tried to analyze and you are now rastering with a bent beam, an area slightly adjacent to your field of view. If you drop your magnification down significantly, you will see your field of view, where you started, walk back to the center of your screen. So, maybe try some coating. If you can, adjust accelerating voltage, if it’s possible drop down lower, turn your current down some more. You can compensate for some of this but you'll eventually get to a point where you can't, and coating then becomes the only way. CHUCK: Is the edge effect really just a result of different saturation curves for the edge versus the flat section? JOE: You've got to your excitation volume, the surface of the portion of it that is generating the secondary electrons that are going to be able to escape.
It’s a bigger area when it’s close to the edge. So, it is the equivalent of tilting the sample, it is just artificial and you instantly got sample tilted. CHUCK: Going back to the topic of dead time, “You mention that dead time needs to be greater than 20%, doesn’t that make pulse pileup increase?” JOE: Not so terribly. It’s a nice 20-25% is a very good to count rate, as an example of dead time, because if you go much lower, your peak to noise ratio is going to become atrocious and your statistics on the measurement become much greater and this is not what you want. Ideally, if you’re doing quantitative work, you know the world was perfect, you like to have all your uncertainties 1% or less. CHUCK: “Is there of recommended minimum or maximum magnification for EDS analysis?” JOE: No, you can analyze at whatever conditions your SEM will be able to be adjusted to. If you have a very large sample, several inches or 6 inches in size, I mean you could do that analysis at 30X magnification And you are just acquiring spectra from the full field of view that you’re rastering over.
If you wanted to get smaller, you just start zooming in and when you increase the magnification in the SEM, what you are actually doing is you're limiting the length of the raster from point A to point B. And then you’re taking the area that the images be projected on and dividing it by the raster length. So, now as you get shorter and shorter in your raster length, your magnification is being increased. CHUCK: Going back again to the topic of dead time, “Are there any disadvantages for a low dead time?” JOE: You just don't have enough signal coming in. You're going to have very poor counting statistics. If you want to have smooth the spectra, good peaks, you're going to have to count for a long time, and the uncertainty is just going to be there and very high.
Can you live with 25 to 50% uncertainty? This is the point where you got to be thinking about is that element even present, is this real? What is the accuracy of this measurement? It is not going to be real good. CHUCK: From Brandon, here, again, “Are there any settings, operating conditions that can be adjusted to affect the sharpness of your EDS peaks?” JOE: Not really affecting sharpness of the peaks, but in the software routines you may have a smoothing effect, and this may or may not be available in all software from all the manufacturers. It just an artificial smoothing out of the peak. If you want to just have it run longer, it will become smoother like that. If you a two second acquisition versus hundred second acquisition the sharpness of the peaks will look a lot different. CHUCK: I think that about does it for the questions, Joe.
Again, thanks for a very informative webinar and we will be doing another webinar here in two weeks, February 4th. Our presenter for that will be Scott Stoeffler of McCrone Associates and Scott's presentation is titled, Identifying Foreign Particulate in Pharmaceutical Products. We hope to see you out here for that in a couple weeks, and thanks again for attending.