SEMs probe beneath the surface

A scanning electron microscope shows surface features, but it also tells you what's inside a sample.

Robert Johnson, Robert Johnson Associates, El Dorado Hills, CA -- Test & Measurement World, 7/1/2001

Figure 1. This electron microscope image of a semiconductor cross section lets you differentiate between bright conductive materials and darker insulators. This view reveals the layered structure and details that you could not see from a top-down view or with a light microscope. Courtesy of FEI.

Figure 2. A commercial SEM looks deceptively simple. The enclosures hide the details of the vacuum pumps, sample chamber, and precision electromechanical components needed to produce images with 100,000X magnification. Courtesy of JEOL.

As the structural complexity of semiconductors increases, so does the need for accurate information that relates various components—and their chemical makeup—to one another. The image shown in Figure 1 provides an electron microscope’s view of a sectioned semiconductor device. Electron microscopes provide high resolution, a large depth of focus, a wide field of view, and easy image interpretation. Measurements of minute features made by electron microscopes play key roles in controlling the processes used to fabricate microelectronic circuits. The image reveals metal, insulator, and semiconductor materials.

In this case, the sample was cross sectioned and prepared specially for side-view imaging, but electron microscopes can also nondestructively “probe” the first few micrometers beneath the surface of a sample to provide information about the material’s composition, its electrical and magnetic properties, and its structure.

You’ll encounter two primary types of electron microscopes: the scanning electron microscope (SEM) and the transmission electron microscope (TEM). Both the TEM and the SEM (Figure 2) use a column of electrostatic and electromagnetic elements to generate electrons, accelerate them toward a sample, and focus them to form an image or a minute scanning beam (Figure 3).

TEMs, as the name implies, produce a direct magnified image of a sample based on electrons transmitted through the sample, similar to how light passed through a transparency projects an image on a screen. In a TEM, a broad beam of high-energy electrons illuminates the entire sample simultaneously. As the electrons pass through the sample, it absorbs, deflects, or diffracts them. A series of magnetic “lenses” focus the transmitted electrons on a fluorescent plate, or other viewing medium, that produces an image of the sample. A TEM can resolve individual atoms in a silicon crystal, so it could measure the thickness of a gate-oxide layer 30 or 40 molecules thick.

Figure 3. A simplified cross section of a SEM shows the source of the electron beam and the magnetic and electrostatic components that focus and scan the beam. As the beam scans a sample, a detector measures secondary radiation emitted from the sample.

But a TEM is more difficult to use than a SEM, and it demands more stringent sample-preparation steps than a SEM. Typically, you’ll find TEMs in research labs rather than in production facilities. SEMs, on the other hand, are by far the more common type of electron microscope used in industrial applications.

Instead of illuminating a sample all at once, a SEM scans a narrow, focused beam of electrons over a sample surface in a raster pattern. At each point in the scan, the electrons penetrate a short distance into the sample and give up their energy as they interact in several ways with the sample’s atoms (Figure 4).

These interactions emit of a variety of radiation, generally other electrons or x-rays. A beam electron—also called a primary electron—may eject a “secondary” electron from one of the outer orbitals of a sample atom. Because secondary electrons can only escape if they originate near the surface of the sample, they reveal a great deal about local topography and provide the highest spatial resolution of all the types of radiation emitted by a sample.

The nuclei in a sample also may scatter beam electrons, which are then called “backscattered” electrons, that a SEM can detect and use to create different types of images. A beam electron also may generate an x-ray or an Auger electron, both of which carry “information” about the composition of a sample. By selectively detecting and measuring these types of radiation, a SEM can provide a user with much information about the properties of a sample.

Figure 4. The electron beam that scans a sample actually penetrates below the sample surface to form a volume that will emit secondary radiation, typically other electrons and x-rays. The resulting radiation can provide much information about the sample’s composition.

Unlike a TEM, a SEM does not produce a real image from the beam electrons. Rather, a SEM constructs a virtual 2-D image by measuring the intensity of each type of radiation emitted from the sample as the electron beam moves across it. The image—built up point by point—represents the intensity of the same radiation type at each scanned point. The contrast between areas of high and low radiation produces a SEM image that includes shadows and highlights. Surfaces oriented perpendicular to the electron detector generally appear brighter than surfaces that tilt away from the detector. SEM images appear almost as if the sample had been lit by visible light.

Know a SEM’s limits

Although a SEM can reveal small features, you still have to know about the limits of resolution. Resolution describes the minimum distance that must separate two features on a sample so a SEM can still view them as separate and distinct features. In a light microscope, the wavelengths of visible light limit resolution to about 0.5 µm. Light at the shortest visible wavelengths can push a microscope’s resolution to about 0.2 µm, which yields a useful magnification limit of about 1000X.

Although a SEM produces electrons at wavelengths several orders of magnitude shorter than visible light, the size of the beam limits resolution. In general, a SEM cannot distinguish two adjacent features separated by a distance smaller than the diameter of the electron beam’s “spot” on the sample.

Many other variables also may affect resolution by changing the volume of the sample in which the sample and the beam interact. These variables include the density and conductivity of the sample, the energy of the beam, the beam-scan speed, the distance from the objective lens to the sample, and the angle of beam incidence. Most SEMs offer resolutions of at least 5 nm, and high-performance SEMs can achieve resolutions that approach 1 nm. As a result, a SEM can magnify features by 100,000 to 200,000 times.

Depth of field goes hand in hand with resolution. In any microscope system, the variables that increase resolution generally act to decrease depth of field, and vice versa. As always, you face tradeoffs. In a SEM, however, the convergence angle of the beam, which determines the variation in spot size as depth changes, is very small. As a result, a SEM offers a much better trade-off and greater depth of field than an optical microscope. The images in Figure 5 show an optical microscope and a SEM image of a sample. Comparing the two images gives you an appreciation for the depth-of-field capability of a SEM.

Figure 5. A comparison of the skeleton of a radiolarian—a one-celled animal common in plankton—examined with a light microscope and a SEM shows the high resolution and great depth of field possible in SEM images. Copyright 1992, Plenum Publishing. Used with permission.

Stereographic pairs of images can produce 3-D views that further enhance the information provided by a SEM. Our perception of depth results from the slightly different angle at which each of our eyes views a scene. A SEM can mimic this effect by slightly altering the viewing angle of the sample between consecutive image scans. By pairing one image with each eye (using specially designed glasses or other apparatus), you perceive depths in the sample. Stereography can let you make quantitative measurements of topographic features in a sample.

Get beneath the surface

Radiation from a sample can also tell you about its composition. The likelihood that a nucleus will backscatter a beam electron depends on the size of the nucleus. Thus, heavier elements backscatter more electrons than do lighter elements. In an image made by measuring backscattered electron intensities, the locations of heavier elements appear brighter than the locations of lighter elements. Because backscattered electrons originate from a large region deep beneath the beam spot, they cannot provide the same spatial resolution as secondary electrons.

The radiation emitted from a sample has even more to reveal. A beam electron can scatter an electron from an atom’s inner orbital. As a result, the atom emits an x-ray whose energy depends on the electronic structure of the atom. The characteristics of such x-rays are well known, so determining the energy of the x-ray leads to the identification of the element that emitted it. The intensity of the x-ray yields a quantitative elemental analysis of the sample. But x-ray images that map element types have relatively low spatial resolution (about 1 µm) because the x-rays originate from a relatively large region and can travel great distances through the sample. Also, these x-rays have low intensities, so images tend to be noisy.

Sample atoms themselves may produce Auger electrons, yet another type of radiation that a SEM can detect to determine elemental composition. But nearby atoms easily scatter Auger electrons, so only those electrons emitted within about 1 nm of a sample surface reach a detector unchanged. Thus, you can use Auger electrons to perform a sensitive analysis of surface composition. Auger-electron analysis places high demands on the vacuum environment in a SEM, and the technique is best performed using a special analytical SEM.

Because secondary electrons have low energy, local electrical fields easily influence their properties. Some measurements can take advantage of this sensitivity to assess the electrical characteristics of a sample. If a sample contains a circuit that includes some positively biased components, the components will attract secondary electrons. So, fewer electrons will escape from the positively biased components, and they will appear darker in the sample image.

Sources offer variety

In a SEM, the type of electron source heavily influences performance and cost. SEMs commonly use three types of sources: tungsten, lanthanum hexaboride, and field emission. These sources vary in their “brightness,” a measure of the current the SEM can focus into a spot of a given size. A high brightness source lets an operator form a smaller spot with the same amount of current, and thus produce images with better resolution. Tungsten provides the lowest brightness, but it costs the least and is the easiest source to maintain. A lanthanum hexaboride source offers a tenfold increase in brightness. Both tungsten and lanthanum hexaboride require heat to emit electrons, but lanthanum hexaboride becomes highly reactive when hot, so it requires a better vacuum to prevent chemical reactions on the source that would decrease its electron emissions.

Field-emission sources produce the “brightest” output of electrons. A high potential applied between the sharp tip of a tungsten crystal and an anode produces a field so strong that electrons “escape” from the tip without the need for heat. Field-emission sources require ultra-clean high-vacuum systems, because any contamination on the sharp emitter point will drastically affect emission characteristics. Although field-emission systems offer the best imaging performance, they’re the most expensive.

All SEMs must maintain a vacuum along the path of the electron beam to prevent molecules in the SEM’s internal atmosphere from scattering the beam electrons. Conventional SEMs typically hold the vacuum level throughout the system at about 10^-5 Torr. Recent developments have led to the appearance of so-called low-vacuum SEMs that allow the sample chamber to operate at pressures as high as a few Torr.

The lower vacuum extends the use of scanning electron microscopy to samples that are too dirty or that outgas too much for conventional SEMs. Some low-vacuum SEMs permit pressures(>5 Torr) in the sample chamber that are high enough to sustain liquid water. Thus, a SEM can examine wet samples in their natural state. (See “Prepare samples with care,” below.) Some of these newer SEMs also incorporate a special secondary-electron detector that amplifies the electron signal and neutralizes any charge that accumulates on a nonconductive sample.

Prices for SEMs vary widely, but you can buy an entry-level system for less than $100,000, although specialized SEMs command prices in excess of $1,000,000. If you don’t want to purchase a SEM, many commercial laboratories offer scanning electron microscopy services. Prices for examining a sample at a service lab depend heavily on sample-preparation needs and the analytical information you require, but they typically start at a few hundred dollars per sample. T&MW

For more information

Goldstein, Joseph I., et al., Scanning Electron Microscopy and X-Ray Microanalysis, 2nd ed., Plenum Press, New York, NY, 1992.

Heinrich, Kurt F.J., Electron Beam X-ray Microanalysis, Van Nostrand Reinhold, New York, NY, 1981.

All you wanted to know about Electron Microscopy..., FEI Co., Hillsboro, OR. www.feico.com.

Robert Johnson heads Robert Johnson Associates, a technical communications firm in northern California. He has held a variety of sales, marketing, and product-development positions within the analytical-instrumentation industry during the last 20 years.