Scanning Electron Microscopy (SEM) Lecture: Principles, Techniques & Applications

All right, welcome. Uh my name is Matthew Hunt. I'm the staff microscopist here at the KNI. Uh this will be the first of three lectures on microscopy over the next few weeks on scanning electron microscopy then gallium focused ion beam and then helium and neon focused ion beam. Okay, so first of all, I want to show an example of what it looks like to use a scanning electron microscope. Here, we're going to scan the beam across a

field of view that's 15 microns wide and we have patterned with metal KNI Caltech and some circles here. So, this is silver. It's got a lot of texture to it. Gold and titanium on silicon. And so, this is what's called the secondary electron image and we'll talk about what that means. The secondary electron image is going to show us topography of the sample. So, as I start the video, and the beam is scanning, about once a

second we're getting a refresh of the image. Here, I'm about to change the voltage on a detector to take us from secondary electrons to what are called backscattered electrons and these are primarily going to give us atomic number contrast. So, you can see the image changed dramatically and now I'm scanning at a slower rate so that now every 5 seconds we're capturing a new image. And then I can go back to scanning a

secondary electron image and now I'm back to more of a topographic image. So, throughout this lecture, I'm going to explain all the different parts of this, the voltage that we're using, the current that we're using, the working distance that we're using, and lots of other tricks that we can employ to take nice images of all kinds of different features. First of all, I'll say all of these resources, including this presentation,

are available on our KNI lab wiki. So, that's lab.kni.caltech.edu. You can search for this quite easily. So, literally everything that you want to find, resources related to microscopes, you can find on our wiki. Uh this is just a slide to show the full functionality. Again, there's a whole page that describes how you might choose a different microscope to use or which microscopes to use for your research in the KNI and this describes all the

functionality. These next three lectures will be going through all these different things. Uh now, you have handouts in front of you, three of them for today that are relevant and then a fourth one and a fifth one for the different focused ion beams. Uh we're not going to go through it right now, but as we go through the presentation, you'll see these little diagrams appear on different slides and you'll have a way to follow along with

your diagrams. For the YouTube uh presentation, I'll put links to them below the description of the video. So, there's SEM, there's a handout on alignments, and then there's also a handout uh considerations for optimizing your SEM images. And we'll go through all of these different things over the course of this presentation. Uh there are also books uh for each of these three types of microscopy that I'll go over. These are the the notable

books that I would recommend. Uh so, uh look at those when you want to uh further study any of these concepts. And what I'm going to try to do over the course of these lectures is show the analogies between the three different types of microscopy. And here's a good place to start. These are the emission sources and we'll see that each of them is a tungsten needle and we'll have some kind of extractor, some kind of suppressor. We'll have an accelerating

voltage setup and uh we'll be able to draw analogies. So, they they all look a little bit different and they have their own attributes that to give emission of electrons or gallium ions or helium ions. They're all fundamentally of the same construction. And so, this will be a a core concept of these lectures, trying to draw analogies across the different microscope platforms. So, those are the the sources, the

column optics. Again, we'll see that they're largely the same. If we look at them in cartoon diagram format, we can see they have similar lenses, they have similar similar stigmators, similar uh source construction, and uh again, that's what we're going to use to ground ourselves and make sure that we understand how uh the analogies work across the different microscope platforms. So, before we start getting into the the

physical concepts, I'll just show some pictures of microscopes that we have in our lab, which look similar to microscopes in all the labs across the world. So, here's an SEM with a gallium focused ion beam. We call it a dual beam system in our lab. So, the SEM is on the main optical axis. The gallium FIB here is offset by 52°. Here's a vacuum screen showing uh that we have things like ion getter pumps pumping on the columns and we have turbo

molecular pump pumping on the chamber. Uh there's other things on here like we have an aperture controller. I'll show what that looks like uh closer up later. We have column isolation valve. That's what keeps our columns isolated from the chamber when we're venting. And I will show you, I think on the next slide, what it looks like inside. So, when we open the door, we vent the whole chamber

in the case here. We we could pull out the door and we see here's our specimen on a stub on a stub holder in an XYZ rotation tilt stage. Uh this is a 6-in stage, our Nova 600. Uh and we have uh the SEM column here. We have a lot of other things inside. The things that are relevant to us today are the Everhart-Thornley detector and the through-the-lens detector. These are secondary electron detectors. I'll go through what it means to create

secondary electrons and detect them and and why you would want to do that. Uh we have other things like gas injection system needles that will be important for gallium focused ion beam presentation. Uh if we look at a reverse view, we see other things. Uh here's the gallium column. Here's what's called an Omniprobe needle. Again, those Uh we have some other things. We have a

magnetic field sensor so we can sense the magnetic field in near the column and then we can cancel that field by running current through loops that are on three different walls of the room. Uh otherwise, we'd get some oscillations in our beam. Um and let's see. Otherwise, some controllers. I think that's Those are the main things today for for SEM. We have a Nova 200. The 200 just means

2-in stage. So, a lot of the same kind of stuff. What's different here is we have a energy dispersive spectroscopy detector and a wavelength dispersive spectroscopy detector. We'll talk about how we can use X-rays to to analyze the uh elemental composition of our samples. Okay, so uh when we talk about lenses and when we get into the column optics, we need to understand the difference between electromagnetic lenses, which are used in electron systems, and

electrostatic lenses, which we'll use in ion beam systems. So, the basic construction of an electromagnetic lens is what we have here. So, you have a a wire coil like copper wire embedded in a iron shell and we run current through that wire so we generate a magnetic field and we allow an opening or a bore in that iron shell. So, this is radially symmetric. So, we leak that magnetic field out into the opening of the or or through into the column. And as an

electron beam passes through, the electrons that are that are off-axis will get bent strongly back towards the main axis because the field is stronger at the perimeter and those in the center won't get deflected at all and those that are say halfway between will get mildly deflected. So, the idea is we want to get to a crossover point or a focus point. Uh we'll talk about electrostatic lenses in the next lectures, but they

ultimately uh accomplish the same task through a different mechanism. It's called an Einzel lens. We have uh positive potential and then two other plates at ground potential. These again, think of them as concentric rings uh or plates. And so, we get we can symmetrically accelerate and decelerate the beam as it passes through the lens. Now, another thing to mention, with the electrons in a magnetic field, they they

spiral. So, this is kind of a very reduced uh cartoon version of what's happening, but in cross-section, you can think of it happening like that. Okay, so that's a condenser lens and that's up in the in the column and I'll talk more about the condenser lens later on. What you're probably more familiar with or more interested in is the objective lens, which is at the bottom of the column. And so, a basic uh objective lens configuration is like

this. So, we have our copper wire coil inside of this more complicated geometry of an ion iron core. And so, what we do is we run current through that and we generate a magnetic field and we allow it to leak out through the bottom. And so, this would become the objective lens plane right at the base of the column. And again, the field will be strongest at the edge and weakest in the middle and allows us to focus the beam down

into a point and then we'll be able to scan that over our surface. The Einzel lens, we just stick one at the bottom of the column and again, we're able to focus the beam down to a point at the surface. This is what we would call field-free mode with the Thermo Fisher FEI systems. Uh we also have a higher resolution mode and we call that immersion mode on our systems. So, you see there's this other coil of wires that's outside of the iron

uh shell. And so, if we run current through through that, we generate a magnetic field that's inside the whole chamber. And if we look at those field lines, they kind of come together between the sample and the bottom of the column. And so, we have what's called a floating objective lens plane, and it's going to be halfway between the column and the sample. And we'll see how that working distance, and we'll describe what that means. We minimize the working

distance, and that allows us to get a smaller probe and get higher resolution. Uh note that when you do this, your if your sample's magnetic, you've magne- uh you've magnetized the pole piece, and you would get attractive your sample might get attracted up to it. So, you don't want to do this with magnetic samples. Okay, so now we've shown the analogies across different types of microscopes. We'll get into the particulars of SEM

here. So, this is the SEM field emission gun. This is the source. Uh so, what we have is a iron or excuse me, a tungsten filament. That's our our cathode. So, we hold that at negative potential. And then we have our anode plate, which we hold at ground potential. And so, say we want a 30 kV accelerating voltage. Uh then we we keep this at -30, this at zero, and we're able to accelerate our electrons uh down

the column. Now, the way that works with a field emission gun is that we do run current through that filament. So, we get some thermionic emission. It's tungsten, and it's hot, so we get thermionic electrons. But we also apply this electric field so that we are able to lower the work function at the tip and get more electrons to emit from that tip. We even coat the tip with zirconium oxide. That further lowers the work function and allows us to get more

electron emission and a higher brightness of our beam. And that's what you're looking for in these systems. Um and what you see here, so I mentioned negative potential, ground, we hold this grid cap at slightly more negative potential, so we get our first crossover of the beam here. Most of those electrons might get collected at the anode plate. The ones that pass through are able to accelerate down the column uh with an energy that's

proportional to the accelerating voltage. As far as the column, we've talked about the source up here. Now, we have a first condenser lens, and usually you'll have two of these in your SEM. Um the whole point is to take what's a fairly large spot here and ultimately demagnify it into something much smaller. So, that virtual source might be on the order of tens of nanometers. We need to get it down to

the order of single nanometers. And so, we have to demagnify, and we do that through the series of lenses. So, you might have two condenser lenses that help you demagnify the beam. Uh so, what they're what the the condenser is doing is it's crossing the beam over somewhere above the the aperture in the system. And then the current that passes through will come down through these octupoles.

So, we have scanning and stigmata octupoles. These are electromagnets, and they allow us to scan the beam over our sample, so we can get a 2D image. And we'll see that the stigmata octupoles also allow you to manipulate the shape of the beam, so you can get a nice circular probe for nice sharp images. Uh we can apply a voltage to this blanking plate, and we can uh measure the current through an ammeter, and it also prevents us from exposing our

sample to electrons when we don't want it to be. Uh and last thing I didn't mention, these gun tilt and shift quadrupoles are used to sort of steer the beam, and I'll show uh examples of of doing that. So, this slide is just to kind of give an overview of the different parameters that are involved. Uh the basic ones you want to keep in mind are the accelerating voltage, the probe current, and the probe convergence angle. And all

of these things conspire to give you a certain probe diameter, which as the operator is what you're ultimately concerned about when you're thinking about the resolution of your images. And there's relevant equations that are involved. We won't go through them all today. On the slides where I go through things like voltage and current, I'll keep the relevant equations there so that we're always reminded that we can dig into the

equations and do some calculations. Uh but these are the the main ones that are involved, and this DP. So, this is the ultimate probe size. It's a function of all these other ones, which you can see up here. So, it's all about calculating the probe size, and that's a function of different aberrations and different uh the parameters like voltage and current and convergence angle, as we said. All right, so probably the most important parameter is the voltage. And

this is usually what you want to decide upon first when you're looking to optimize your image. So, the voltage controls the energy of electrons in your system. We on a typical SEM will have as high as 30 kV accelerating voltage and as low as maybe a half a kV or maybe 0.1 kV or even lower. And so, if we have 30 kV, then on average each electron has 30 keV in energy. 15 is 15 keV, five is five, okay? Um and so, the higher the energy, when the beam makes contact with

your sample, the the deeper it'll penetrate and scatter. And so, we refer to this as an interaction volume. You might also hear it called an excitation volume. And so, you can see that interaction volume gets smaller at lower voltage, and this is simulated for silicon. Uh there's a scale bar here, 1 micron. So, at 30 kV in silicon, we're looking at about 8 microns deep. This is like 2 and 1/2 at 15, and only a couple hundred

nanometers at five. Uh now, we see that the interaction volume is broken up into different colors, and we'll talk about all those different uh interactions as we proceed. Now, um when we want to image, we would want the highest voltage should theoretically give us the highest resolution because for one thing, high energy, if we look here, higher energy gives us a smaller

wavelength, and therefore we can theoretically form a smaller probe. But we're not really wavelength limited in an SEM. We're really limited by aberrations in the system. And it turns out that spherical aberrations and chromatic aberrations are worse at lower voltage. And by spherical aberrations, I mean the imperfections in the lens. So, if if you have imperfections in your lens, then the farther out your

your electrons go, the more subject they are to those aberrations. And also, the longer time they spend in the lens, the more subject they are to aberrations. So, higher voltage, the electrons have higher energy. They're literally traveling faster through the lens, so they're not going to be affected by these aberrations as much. Chromatic aberrations are the range of energy spread of electrons coming from the tip, and we'll get a smaller energy

spread at higher voltage, and therefore all of your electrons will be closer to each other in wavelength as they pass through the lens, and therefore can get focused down better into a small probe. So, it's really about aberrations, and the aberrations are are not as bad at higher voltage, so we can get higher resolution there. Now, with a that's true for conductive specimens. So, here this is just gold on carbon. We have a much sharper image at 30 kV

versus 15 versus five. Uh but if you have a non-conductive specimen, that doesn't really help you because putting all of this electron uh negative charge into your sample, you can build up charge. Your beam will start deflecting. You're not going to get a nice image. You'll get a lot of artifacts. What people often call charging artifacts. So, if we have a non-conductive specimen like this, it actually becomes more

useful to image at lower voltage. Because at lower voltage, we have more secondary electron yield, and we'll talk about secondary electrons. Uh and so, our current in versus current out, we can actually get to an equilibrium uh by going to a progressively lower voltage. So, like here, we have just as much current going out as signal as going in, and therefore we're not

worrying about accumulating charge, uh and we can get artifact artifact-free images. Another thing to think about, at 30 kV, if the interaction volume is literally larger than the features you're trying to image, then electron secondary electrons will spill out all different surfaces, and your sample will end up looking translucent. So, if you have small features like this, like these small beams, the low voltage

is good if it's non-conductive, as this is. It's uh hydroxyapatite, and the interaction volume is now smaller than the features, and therefore we're confining our secondary electron emission to only the surface. We're not getting it spilling out of other surfaces. All right, so now we have our SEM current, and uh one way to think about current is that it controls the the spot size. So, all everything controls spot

size, as I said, or or the probe size. But here, we have some um some numbers from an old JEOL presentation. At 22 picoamps, you might get a 2 nanometer probe size. That's the best about the best you're going to do in an SEM. And then 400 picoamps, maybe 3 nanometers, and then it starts to get larger, say 10 nanometers or up around a nanoamp. And of course, this isn't we don't have a spot like this. There's like a Gaussian profile. It's just about half of the

probe current in in that uh diameter. And if we look at the images formed with these different currents, we'll see that we have sharper features at low current with a smaller probe size and a more blurry image when we have a larger probe size here. Um and so, what we want to think about when we do when we're changing the current is while we when we lower current, we'll get a smaller probe diameter for better

resolution, but we'll also get less signal. So, you're really balancing your probe diameter and therefore your resolution versus the amount of signal that you get. And I have a tip here that if you're trying to figure out which current to use, look at the contrast value on your system. So, if your contrast is really high, like 95 98%, then you you can't afford to go to lower current because you won't be able to to maximize the the grayscale range

of your image. If anything, you'd want to have more current so that you can you can spread and get better contrast. Uh but if your if your contrast values in the 50s 60s 70s, that tells you you have more current than you probably need or more signal than you need and you can lower the current and therefore increase the resolution. All right, so how do we control current? We do that through apertures and we do it through the condenser lens or

condenser lenses if you have more than one. So, aperture's pretty simple. We can just vary the the physical size of the aperture. Smaller apertures let through fewer electrons. Uh but it's not just that. It's that we let through electrons that are more on axis. And you want on axis electrons because when they enter the objective lens, we'd like them to be closer to the center of the lens so they can come to a point. If we have

a larger aperture, we'll get uh electrons that are spread over a a larger range and it's going to be harder to get them to focus back down to a small probe and that's why you get worse resolution with a higher aperture or a larger aperture. All right, so here's our aperture wheels. We can just physically select them on our microscope. Sometimes you'll have an automatic aperture changer. It's

all the same. Um but the the next thing we think about is the condenser lens. So, what the condenser does is it's going to change where this crossover point happens inside of the column. So, if I go here, we have a unitless uh metric called spot. So, say spot seven on our microscopes is really high current. So, we're crossing the beam over basically here. I'm putting it at the aperture plane, which means we take

all of this current and we slide it through the aperture and it's going down to your sample. If we increase the strength of our condenser lens, then we force these electrons to cross over higher in the column. So, now we have a spread beam illuminating our aperture and we'll get fewer electrons through. And then if we increase the condenser further, then it spreads more and we get fewer electrons.

So, usually you'll have like seven different condenser settings. So, if you think whatever size your aperture is, this might be spot 1 2 3 4 5 6 7. So, a more condensed beam, you'll get more current through. A less condensed beam, you'll get less. And again, the angle of those electrons will matter because if you're more condensed, then you're going to get all these through and then as they get down the column, they'll have to get brought

back together. But if you spread it out wide, you're only going to get these on axis electrons through your aperture. Uh and here we see some of those relationships that I was talking about in terms of condenser lens strength to probe diameter resolution and signal. Okay, so So, this is an interaction volume and we're going to build it out and learn about all the different signals that are involved. So, the first signal and the

primary signal that most people think about and care about is the secondary electron signal. So, if we have a primary electron from the beam, it's going to inter- interact inelastically with the electrons in your atom. Those electrons will get kicked out and we call them secondary electrons. So, they're low energy, uh usually 50 electron volts. Um and it's not and I have it colored here blue at the surface. This is where they're coming

from. The truth is we generate them everywhere, but the only ones with enough energy to escape are in this top part of the interaction volume. So, they're able only the top uh 5 10 nanometers can we get emission and therefore when we collect our image, we're we're taking signal from the top 5 10 nanometers. Therefore, we're going to see the topography of the image. And so, we have something like an Everhart Thornley detector and where we

put a positive bias on it, we attract the electrons to it and that's how we end up counting the electrons at each pixel. More electrons at a pixel is a brighter pixel. Fewer electrons is a darker pixel. If we put, say, a thousand by a thousand of those together, we get a nice 2D image. Uh we also have usually a detector inside the column on these FEI systems. We call it a through the lens detector. Other systems call it something

different, but we can put a positive bias on that and we can attract secondary electrons up into the column and detect them that way. Uh this is has will have directional shadowing and this will have sort of radially symmetric shadowing. And then we'll see this is also used to do the highest resolution imaging on our systems. Next, we can talk about backscattered electrons and so those are electrons

that are elastically scattered by the atoms in your specimen. And so, because they're elastically scattered, they lose less energy. So, they usually emerge with, say, 50 to 100% of their original energy. So, say you had 30 keV uh electrons from a 30 kV beam, they're going to emerge with something like 15 to 30 keV, which means they can emerge from deeper in the specimen. So, they're not going to be very good at showing you

the surface topography because they're not literally coming from the surface, but rather they're going to show you atomic number contrast because the elastic scattering is a function of the atomic number. So, a gold atom will elastically scatter more electrons than, say, a silicon and therefore your gold will appear brighter on the screen if you have a backscattered electron detector there to collect that signal. Now, one thing I do want to point out.

So, this is a really nice Z contrast image of gold and titanium and silver on silicon and they're all bright according to their order of Z. Uh we also can kind of see some pretty good Z contrast here with gold on silicon. So, it is a little bit easier to to emit or uh a secondary electron from a higher Z element like gold than it is from silicon. So, you do have some atomic number contrast just through secondary electrons. Uh but if you want a pure

a pure Z contrast image, you want to use a backscattered electron detector. And then lastly here, we generate characteristic x-rays. So, if we have a primary electron knock out an inner shell electron, an outer shell electron will drop from high to low energy and when it does so, it'll emit a photon of equivalent energy and that photon is on the order of an x-ray. And so, every element has its own energy transitions and therefore emits its own

characteristic x-rays. And you actually emit them from the whole volume, but the characteristic ones usually come from this spike shape and that's something that we can see when we do uh Monte Carlo simulations of beam specimen interactions and I I show that in a different video on YouTube. I go through how you can simulate the interaction volumes for your materials and show exactly why I draw it like a spike in

that way. But the x-rays that are in line of sight to the x-ray detector are the ones that get counted and they get uh used to generate a spectrum that will show you which elements are in your specimen. And here we do mapping. So, we've taken that and we've broken it up into its silicon gold silver and titanium maps and we can see exactly where they all come from. All right, and so on your handout, I

also have a table like I have here and it gives you uh some guidelines on which detectors you're going to use, uh what information you're going to detect with them and about what the recommended current range that you will need. So, we generate a lot of secondary electrons, so we don't need a lot of current to to get a good signal to noise image, which is good cuz you can operate at low current, which means high resolution for your topographic image. You need more

current for backscattered electrons because the elastic events are more rare. So, you need to really increase your current to get enough signal. Um and then you won't have as good resolution in that. And then you need much more current usually for x-rays because while we do generate a lot of them, only these on this small solid angle, this line of sight to the detector, those are the only ones that we count. So, you do need to be in the

nanoamp range typically. All right, so just showing these uh in larger magnification here, you can see a secondary electron image. You see all the topography, all these imperfections. When we go to the backscatter image, the image flattens out a little bit. You can still see some of this, but it's it's much flatter, I would say. Uh but better Z contrast. And in our

our EDS maps, we we can see the elemental contrast. Okay. Uh now, one thing I I showed in those little data bars, I had different voltages used for the different collecting different signals and I'm going to show why you might choose a different voltage to capture these different signals. So we have our um our interaction volume here for 30 kV and this is what it looks like in the

simulation software. This is Casino Monte Carlo software. You can download that. It's free and and you can use it. Uh so that's what it will look like when you just run the simulation. These blue paths are reabsorbed electrons in the red paths are backscattered electrons. So that's kind of how I'm drawing this the the overall shape of the image and where I'm getting the backscatter volume. We can also look at the energy by

position map and we can see that the electrons have the most energy where they enter here, sort of like a hot spot and then as the electrons scatter they lose energy and therefore can't create as many uh characteristic x-rays. So uh when you're forming your your spike for your x-rays if you're going to get most of them up here and it's going to trail off as it goes deeper in the specimen. Okay, so when I did those images I did

chose 30 kV, so the highest voltage, a low current, 50 picoamps and a small working distance and I'll talk about working distance but that's going to give me the best resolution for this topographic image. And even though the interaction volume is really large, the only ones that we actually use for the image, like I told you, are the ones from the top 5 10 nanometers. So it doesn't matter about dumping all that charge into our sample

as long as it's conductive. For the backscattered electron image, these are metal films and they're only about I think 20 30 nanometers thick. And so if I want to I have gold on titanium here. If I want this to look like gold and not like the titanium that's beneath it, I need to keep the backscattered volume confined to the the top layer of gold. So at 30 kV or excuse me, 3 kV, the entire interaction volume is in the titanium

and especially the backscatter volume is in titanium. So you don't see any contribution to the gold from the titanium below it. And so that's 3 kV. It's a higher current, about an order of magnitude higher, 500 picoamps and I used a different working distance. And again, I'll talk more about that. Here for x-rays we need to know where that spike is going to be or where the hot spot will be. So it's going to be

about here. Um but we So you see some of this electron scattering through. Uh we need to have a high enough energy so that we can excite x-rays from all the different elements. So the highest energy was silver, 2.96 keV. So you typically want your voltage to be about two times higher than your highest x-ray that you want to to capture. So probably should have went to 6 kV but I think I was keeping it at 5 to try to confine

that interaction volume as close to the surface as possible. But the general rule of thumb, as I say down here, two times the voltage of the highest uh energy that you're trying to detect. Okay, and so here's a slide on energy and wavelength dispersive spectroscopy. So in EDS you're just getting all of your different x-rays, all the different energies to the detector all at once so you can form uh a spectrum like get you get all your elements at

one time. And then in wavelength dispersive spectroscopy we have an an analytical crystal, usually several of these and what we do is we sweep an energy range. So we can change the crystal, we can change its angle and what we ultimately do is if we're trying to filter out and say how many counts or how many x-rays we have at a certain energy, only those x-rays that satisfy the Bragg condition and diffract and hit the detector will

get counted. So we slowly cycle through all the energy and therefore we're able to um get ultimately what is a higher resolution spectrum. So here the outline is EDS. It's going to be a broader spectrum and the WDS will be much sharper. So you can get better quantification with WDS but it's a slower technique. All right, so now we can talk about working distance.

This is uh literally the distance from the objective lens to wherever the focal plane is. And so as we change the strength of that lens by changing how much current we flow through that current loop, we're changing the magnetic magnetic field and therefore changing where this beam crosses over as a crossover point or a focal point.

Uh so practically what we usually do is we get the image in focus. A sharp image means we're scanning a small probe over the sample. And then we tell the microscope we hey, this is where our sample is. So if this is telling me I'm 8 mm away, we click a button and that becomes the coordinate system for the microscope. It knows now your sample is 8 mm away and it can navigate according to that. So that's the practical side and if you

then change focus your Z on your microscope will stay the same. It knows your sample is still there but you're just moving your focal plane up and down. The theoretical side is that the convergence angle of the beam will change with the working distance. So if we have a short working distance, we have a steep convergence angle and the equations would show us that we're going to get a smaller probe size.

And if we have a longer working distance, we have a a more parallel beam a a shallow convergence angle and we're going to get uh poorer resolution from the larger spot size. If we look at these two images, uh the other side of working distance is the depth of field. So we have an inverse relationship between the resolution and the depth of field. So if we have this long working distance we don't have great resolution but we'll

have really great depth of field. So these are two pretty large trusses from the Greer group here and they're they're tilted and they're a couple hundred microns apart but we're seeing them both in focus at the same time. I put the focal plane somewhere in between them and I can see them both with similar resolution. Whereas at short working distance, if I do the same thing, so this is 6 mm versus 24 then I can only get this one in focus and the one in the

back falls out of focus. So think of your working distance as a balance between depth of field and resolution. We also can change uh well here's the resolution. So at shorter working distance we have sharper features here and at longer working distance we have a little more wiggles in the beam so we have poorer resolution. The spot size or the probe size is also larger. Uh but as this tip says, practically

speaking, you're probably going to be confined geometrically by something else. If you're doing gallium focused ion beam, we'll see that you're going to be at what's called eucentric height and if you're doing EDS, you're going to be at the optimum height where you get the most x-rays to the detector. So this is really this balance of resolution depth of field is really when you are just trying to do imaging only. Uh I mentioned eucentric height. That is

a height in the chamber where at any tilt angle we'll be imaging the same thing and we'll see it's very important for us to be able to image what we cut at the same time at any different angle. So here I tilt to 52° and back to zero and we can see that this corner that I'm looking at is in the same spot. So you always most of the time you want to do your work at eucentric height. Uh it's a good way to have a a geometrically reproducible spot in your chamber. So

from week to week to week your sample is always in the same place and you can compare all of your images exactly to one another. Uh and then we'll talk more about this as it relates to using gallium focused ion beam in the next lecture. All right, so I just want to show a couple videos through uh the next couple minutes. So

if we we can watch this happen. Hopefully. All right, so here I'm tilted to 0° and we can see this KNI Caltech logo. As we start to increase the tilt angle ideally this wouldn't be moving at all. If it didn't move at all, we would be at eucentric height. If it moves a little bit, that just means we need to make a

slight adjustment to the Z. And so once we get there to 45°, I'm just doing a little thing where I'm lowering the height of the of the sample. Now it's back to where I started from and now if I tilt back to 0°, it's going to stay stationary in the middle of the screen. So that means I'm at eucentric height and your microscope is set to be safe at eucentric height. So if you're at high tilt, you should be safe even if

it looks unsafe. Okay. Um let's see. All right, so there are different imaging modes. So I talked about field free mode where we just have that single objective lens. Uh this is what gold nanoparticles might look like at 5 kV and in low current. This is what they would look like with the immersion lens uh where we are as I said having a smaller effective working distance. So

here's our objective lens plane that's giving us this image. Now our objective lens plane is floating and it's closer to the sample. So while our our physical working distance didn't change, the effective working distance of the beam has changed cuz we it's smaller here, so we can get a steeper convergence angle and get the benefits of higher resolution from that. Uh what this does though, when we have this this magnetic field here, we can't get electrons out

to the Everhart-Thornley detector, but we can funnel them very efficiently up to the TLD that's inside the column, and we're going to get high signal uh in in addition to the the small probe size. So, that's how we do uh immersion mode imaging. Then you also have something called weak immersion, or also called EDS or EDX mode. And the purpose of this is it kind of gives us resolution in between these two, but but more what

it's there for is to trap these electrons, funnel them up, so you don't have electrons hitting your EDS detector and registering as as x-rays on the detector. That would give you a big background on your x-ray spectrum. Uh we also have something called scanning filters. So, once we've selected voltage, current, working distance, we might end up having some imaging artifacts like we see here. So, this is one frame where we put the beam

at 25.6 microseconds at each pixel, and we're going to get some horizontal scanning artifacts. Whereas if we scan the same frame 256 times at 0.1 microseconds, and we we average all of those frames together, we get a nice uh smooth image. And so, what's happening is if the beam is dwelling at a long time at a non-conductive material, it builds up a little charge, the beam deflects, we get some artifacts

associated with that. Whereas if we scan really fast, there's not enough time to build up charge on any one frame, so every frame is artifact free, but it's low signal. So, we have to average a bunch of these frames together, and that's how we get the high signal and no artifacts in our image. So, we have frame integration, frame averaging, and they're kind of same uh they're just slightly different.

Averaging is like a moving average, integration will just do the the total number that you ask for. And then with our some systems, a lot of SEMs now have line averaging where you average the same line a bunch of times before going to the next. We have that on our helium microscope, and I'll talk about that in the helium lecture. Uh this is a video that demonstrates this. So, there's no artifacts here, but if I

have a longer dwell time, so it a longer frame time, cuz each pixel is dwelling longer, we get these sort of horizontal artifacts. Uh something you can do to test if the artifact is geometric, or if it's real, if it's like a real something on your system, you can rotate the image. So, just rotate the axis of the scanning axis of the beam, and sometimes you won't get as much many artifacts when

you say have these squares that are rotated. Uh so, so that told me I this is an artifact, and now I can try to fix it by doing frame averaging. So, I'm taking a 50-nanosecond dwell time, averaging it say 256 times, and getting uh nearly artifact free image. Okay. Uh now, on the other side of your

handout, we have alignments. So, we're not going to go through Well, we'll we'll do a couple small things here on alignments, but the basic workflow is you always want to adjust the source tilt, so get the beam concentric down the column, and then we'll use our focus to tell us if we have an astigmatism problem, and to tell us if we have what's called a lens alignment problem, and then we'll fix those things according to what I like to call the

visual cues uh in your image. Uh so, why would we have uh something like astigmatism, which will give you stretched features in your image? Well, if you have astigmatism, meaning your beam isn't quite circular, it's elliptical, then it'll be elliptical above the focal plane, and then elliptical below the focal plane rotated 90°, but you'll always will have a circular

beam profile or cross-section at the focal plane. And then what you're going to do is use these uh these stigmators, these electromagnets to to try to reshape your beam into being circular. So, here's a video that demonstrates what happens where if I focus to the right, then I get this stretching. So, I've I've taken that ellipse uh that's below the focal plane, and I've contacted it with the surface.

And so, here it's I'm getting all these features stretching. Well, here, let me back up a little bit. Sorry. So, here I've I'm contacting the sample with an ellipse that's say below the focal plane, and then if I move the focus to the other side, then I'll get that ellipse to rotate 90°, and now it's it's going to everything will be stretched that way. So, what you want to

do is adjust your focus so that you're halfway in between where it's equally blurry on all sides, and then you will adjust the stigmator to actually tighten up the the change the the beam from being elliptical to circular, and then you get a nice sharp image. And so, there are uh videos like this on YouTube where I I talk my way through it, and you can follow those to to learn

more how to do the alignments. Uh this one is what's called the lens alignment. And so, if you move your focus and your image is shifting, that means your beam is not centered on the objective lens. Uh and so, it because your beam is say if it's on the if it's off-center, when you change the strength of the lens, it'll cause the beam to deflect. So, what we do is we wobble the image, which just means

wobbling the focus, and then we use one of these deflectors, these quadrupoles in the column, to get the image or or the beam to be centered on the objective lens. And now when we change focus, it's just goes in and out of focus and looks like it blinks in place. So, those are the two main adjustments you'll make, astigmatism and objective lens alignment. And so again, all these are you can find on YouTube, and you can watch them

there. Uh now, there's another thing that another control we have, the detector bias. So, um we have our Everhart-Thornley detector and our through-the-lens detector. If we put a positive bias on the Everhart-Thornley detector, we'll bring in our secondary electrons, maybe a couple backscatter electrons will get

through, but it's mostly secondary electrons. We can put a positive bias on the TLD, we'll get those secondary electrons, we'll also get some backscattered electrons, but it's going to be primarily a topographic image. We can also put a negative bias on the TLD, which suppresses our secondary electrons and only lets through backscattered electrons, and that will give you more of a Z contrast image.

Uh we can put a stronger positive bias on the TLD and and bring up a lot of secondary electrons, but for some samples, like if you have an insulating sample and you're at low voltage, that can actually take too much signal off of your sample, and it can leave it locally positive, which will then give you some imaging artifacts. And so, what we can do in that case is have a 0-V bias on the TLD

and allow some electrons to go up and some to go back down to the sample, and we can get charge equilibrium in that manner. So, here's a video that shows how we can adjust the the voltage on what we call the suction tube of the TLD. So, I go to backscattered electron mode, it's a -50-V bias, you see I lost a lot of my signal because I'm suppressing a lot of it, uh but I but the signal that

I get through is going to be show me a flatter Z contrast image. If I go back to secondary electrons, I get my signal back, and it's more 3D or topographic. Okay? So, you can see how much more three-dimensional that looks. And then if we go to uh say a custom mode, I can actually slide that suction tube voltage, and when I go to a higher bias, we'll get more signal, so you can use that to

increase your signal. And when I go negative, we lose it all, and we just have to adjust our contrast to get our our Z contrast image back. All right, so that kind of wraps up all these conditions or the different parameters in that you have these considerations for optimum SEM imaging part of your handout. So, I'll go through a couple other things here before showing some other examples and

applications. So, we have a measurement calibration standard in our lab. So, it's a NIST derivative standard. So, on there we have lines of different spacing, so if you want to really um check the measurement of your image, you can't really trust that everything is 100% accurate. These microscopes are usually dialed in to like plus or minus 1 to 2% in terms of their length calibration. So, you'd want to whatever image settings you use to image your

sample, take those exact image settings, image the the standard at the right size, and then just do a comparison. So, here for instance, uh I was trying to measure something that was on the order of 10 nanometers, these lines, and so, we looked at 100 nanometer center-to-center distances and just did the easy calculation to figure out that, you know, I think we were off by just a couple percent, but you can then do the adjustment and get the real

measurement value of your features. Uh something else we have in the lab, sample preparation. I mentioned Well, if your sample is not conductive, then you can go to lower voltage and I usually train people to do that because people don't often want to coat their devices with a conductive material because they want to still make measurements on them. Uh but, if you're able to, you can coat your specimen

with, say, carbon. We have a carbon evaporator. And this puts down like 5 nanometers of amorphous carbon. You won't see it in your image at all, even if you're imaging with um at the highest magnifications. Um and it's it's really conductive and works really well. Uh then we have a uh oxygen and argon plasma cleaner in the lab. So, if you ever do SEM and you get these black boxes, then what's happening is your

if you have any carbonaceous material on your sample, then as the beam hits it, you're locally heating up that area, which evolves off these carbonaceous gases. And then, because we have secondary electrons flying around, we actually can create a plasma locally. And we can can get chemical vapor deposition of those carbonaceous species back down onto the surface of your sample. And now they're more like localized at the surface. And carbon

doesn't have a high secondary electron yield, so it ends up looking darker. So, you get carbon coating your sample, and then you get these these black boxes. So, what you could do is put your sample in an oxygen cleaner, oxygen plasma, remove the carbon, and therefore, when you go and make your images, you won't have these black boxes. And you should be able to actually remove this carbon with that oxygen plasma if you want to recover the uh

functionality of your device after imaging it. Okay, and these are details of that plasma cleaner, which we won't go through today. All right, so here are some notable applications and techniques. Uh I mentioned these e-beam simulations. So, a lot of people in our lab, especially, are doing e-beam lithography uh to to pattern the the different parts of their devices. And so, when you do

that on our electron beam pattern generators, they're 100 kV. So, at 100 kV, we know our electrons have 100 keV, and they're actually going to penetrate about 80 microns into silicon. Uh whereas we talked about with 30 kV in an SEM, it's about 8 microns. So, you have a full order of magnitude difference there. But, when you're patterning something like a resist layer on the top of your device, what matters is not really how

deep the electrons go, but what's happening at the surface. And so, if we have even 500 nanometers of polymethyl methacrylate, PMMA, then if if we want to know what the resolution will be as we pattern it, we want to look at what's happening up there. So, at 100 kV, those electrons are moving faster, we can get a smaller probe size, and they're basically slicing through that PMMA and defining a very narrow path through it. And that's how you're going to get good

resolution on your resist. At 30 kV, your electrons already start scattering as they go through that 500 nanometers, and that's going to mean you're going to uh pattern a a wider spot per per unit pixel that you're exposing. Now, that said, you can still do really good lithography at 30 kV. We have an SEM that has a pattern generator on it, and I've used it to pattern 20 nanometer lines. This is titanium lift-off. So, we did a we did a patterning step,

deposited titanium, lifted off the resist, and we're left with a nice arbitrary pattern like this. So, you can do that. You can also do dots, and I've done as small as, say, 17 nanometers, again, lifting off uh a a very thin metal sample. And this requires using a really thin resist uh layer, which you can't always use when you're doing your devices, but regardless, you can get high-resolution

patterning. We can also do uh alignment. So, I've been using this logo throughout the presentation. So, this was a three-step process where we went from our CAD, we did titanium lift-off with alignment markers, and then we used a backscatter electron detector. So, now this is all coded with resist. So, how are you going to see your markers through the resist layer? You

can't really use topographic signal cuz it's all they're all buried. But, you use the backscattered electrons that come through, and we get Z contrast between the titanium and the silicon, and therefore, we can see our markers, align to them, and then we can write the next step. So, this was gold KNI Caltech with its own set of alignment markers, and then ultimately, we align to those and did silver for those dots. This is our old logo.

So, you can do these multi-step patterns uh in device construction in our uh 30 kV SEM system. We also have an environmental SEM in the lab. And what's unique about that is we can actually operate at a higher pressure in the chamber using this differential pumping arm to prevent our uh to prevent gas from traveling up the column and oxidizing the tip. And we have a couple of different detectors.

And what it's really useful for are two things. One, imaging biological specimens, where we don't want them to dry out at at high vacuum. Uh so, we can maintain their morphology with a higher pressure, so something on the order of uh of single millibar or or 10 We can go from 0.1 to 27 millibar, whereas our vacuum in the in the high vacuum system is on the order of e minus five, e minus six millibar. So, we're about five, six

orders of magnitude higher. And these are This is a artificial plasma membrane, and we're seeing even down to surface texture on the order of of single nanometers. The other thing we can use it for is to image really non-conductive specimens. So, when you the way that you have the high pressure is you have water vapor in the chamber. And so, that water vapor is pretty good at wicking away charge that would otherwise accumulate on your

specimen. So, uh if you have a highly non-conductive specimen, you put it in at high pressure with the water vapor, and you can image it with um without coating it with anything. So, this was done, I think, uh with all the tricks that I could pull on the one of our immersion mode systems. And this was done with fairly little effort on the environmental SEM. So, you get a little bit different contrast, but

ultimately, we're imaging the same kinds of features on this 3-5 material on glass. We have an electrical probe station in the lab, so it's a four-point probe station. It fits over the stage. We're able to guide the probes to our different contact pads and do various measurements depending on what kind of uh function generator that you hook up to the system. We have a hot and a cold stage, so we

can both heat up the stage, I think, to uh 240 C, and we can use liquid nitrogen to run nitrogen gas through it like a heat exchanger, cools down the gas, that gas flows through the stage, and it can cool it down to -185 C. So, you can use this as a way to image your specimen at higher low temperature. All right, some other uh techniques, and then we're about to wrap up. Uh so, one thing I don't have a full presentation on atomic force

microscopy, AFM, but it's a really good technique to use to to confirm what you're seeing with an SEM. So, an SEM, we talked about it gives you topographic contrast, things look 3D in nature, but it doesn't tell you quantitatively how 3D in nature. You can tilt your sample, and you can actually make a measurement in cross-section, and you can you can be pretty confident with that. But, if you're just looking at um say, like the profile of your nanoparticles or

something, you're not going to be able to extract quantitative topography data from it. So, an AFM is a uh stylus, and we usually operate it in an in a tapping mode, and you kind of tap it physically along your sample, and it's able to very accurately measure the Z topography. So, this is a great way, like I said, to to confirm what you're seeing with SEM. Um there's lots of different modes. It's not just topography. You can measure things like uh the magnetic field as a

function of 2D space, electric field. Um you can do uh surface potential, you can do piezoelectric kind of measurements. So, there's all sorts of types of scanning probe measurements you can do. Our particular AFM has something called peak force tapping mode, and it allows you to generate a force curve for every tap. So, we we oscillate at 2 kHz, so we can generate 2,000 force curves first per second, or a force curve for every

pixel that you're scanning. And then, by going in and looking at the shapes of these curves and the different uh components of it, you can end up extracting mechanical property data from it. So, you can even figure out sort of qualitatively quantitative information about the elastic modulus, for instance, or the adhesion forces that are there. It does take some calibration in order to really call it quantitative, but that is something that is available in our

lab. I also don't have a full presentation on transmission electron microscopy, our TEMs, but we do have two TEMs in our lab, a 200 kV system. So, if we want really high resolution, we talked about going higher in voltage to get higher energy electrons, smaller wavelength. So, now we can have 200 keV and on this one we can have 300 keV. And so, we're going to go progressively higher in resolution and be able to do

atomic scale imaging. So, all the different parameters are here. You can look up the presentation at home and we have, like I said, on the wiki, you can go and look up all the different materials we have for these TEMs. Some examples here are imaging quantum dots in what's called high-resolution TEM mode, where you can see lattice planes of these quantum dots. We did some bright field imaging of metal

insulator metal insulator metal devices. And this was a grain boundary study to kind of see when we do an evaporation of, I believe that was aluminum, what kind of twin boundaries do we have, what kind of grain boundaries do we have, what is the the grain size that we have. This is not information that's readily available in an SEM. Especially, this is like, you 50 60 nanometers, about the thinnest we can do in an SEM in cross-section. We've

we've seen 10 nanometers, but you're not going to really get that kind of contrast with, you know, grain boundaries and twin boundaries. This is a quantum cascade laser, so molecular molecular beam epitaxy of all these different layers as small as 1 nanometer. So, here we're seeing in cross-section what that quantum cascade layer looks like in the TEM. And then, of course, you could do diffraction studies

in a TEM, where you can analyze the the crystallinity of your sample, what crystal planes you have, what the spacing is of your lattice, and that's all done through electron diffraction. And what I'll go through in the next lecture on gallium FIB and a little bit later on helium FIB is how we can make these samples. So, we we want to extract a very thin sample. That's how you do TEM. You need

a very thin sample, usually 100 nanometers or thinner, and we can use the gallium focused ion beam to physically cut it free and extract it from the sample. So, in the next lecture, I'll talk about how we can use a omniprobe, a tungsten needle, to take a a thin sample out of our bulk material that we've cut away with the gallium and how we can weld it to the grid with platinum, again using the gallium beam, so that we can thin it down to electron

transparency in a TEM. So, get it less than 100 nanometers. And then, in that lecture, I'll go through all these different examples of gallium FIB, how we can make we can deposit material, make devices. This is a silly one, but you can make real functional devices by patterning directly with gallium. We'll do cross-section. I'll show examples of cross-sections. Here's a 7-nanometer layer that we were able to

image in cross-section. This is what a gas injection needle looks like that dispenses the the gas, the precursor gas that you use to make a deposition. And we'll do some other things like how do you automate your microscope in order to to cut patterns, say say overnight. And then, the third lecture, I'll go through examples of how we we do with the helium and neon focused ion beam.

So, we can get really excellent imaging. I'll talk about why we can get better imaging with a helium beam than an electron beam. How we can image non-conductive materials by balancing the positive charge of helium with an electron flood gun. We can do cross-sections because we have a gallium FIB attached to our helium and neon neon system. And then, I'll show how we can pattern

things like graphene nanoribbons, get high aspect ratio patterns, and do other kinds of device fabrication at smaller scales than are available with a gallium FIB or sometimes even with e-beam lithography. Okay. So, that takes me to the end. You have your handouts. I showed them at the beginning. You have your SEM handout. You have your FIB handout. That will be for the next lecture.

And we have the alignments handout. And I believe that's it. So, with that, we'll end the lecture here and I'll take any questions that you have. Questions? So, I know I just hit you out with a hammer for like an hour. And and truthfully, like I designed this presentation so that when people ask me questions, I send them the like couple slides, you know, like in the slides when you

download them, they have slide notes attached to them. So, if you had a question about environmental SEM, I'd send you the three slides on that and you can read about it. But, this is just meant to be a resource for you all to to download and use depending on what type of microscopy you're using, but it's I think important to understand these physical concepts that are involved and that's why we spend a lot of time going through

this. Because if you can understand what's going on on this one piece of paper, then that's about 95% of of the game in terms of being able to optimize your microscopy to get the best results in an SEM. All right. So, if there are questions, you guys can stick around and ask. Maybe we'll we'll use this as an opportunity to end the video recording and we'll see you next week if you want

to return for gallium FIB.