Scanning Electron Microcscopy (SEM)

History

The first optical microscope was invented in 1590 with many improvements in design and lenses made over the years.  Current models allow scientists to ‘see’ specimens down to ~0.2um with best magnifications possible typically between 500 and 1000x, well beyond that capable of a human eye (~0.2mm).  The smaller the wavelength of light used, the smaller the object that can be observed.  Since the visible light wavelength range is 380-740nm, optical microscopy can’t be improved to achieve greater resolving power.  Thus, scientists began looking for other ways to provide higher magnification of specimens.

​

In 1929 Louis de Broglie published a paper claiming that particles can behave like waves.  Thus, electrons (which are sub-atomic particles) should have wavelike characteristics with wavelengths much smaller than visible light.  Around the same time researchers found that electric and magnetic fields could shape electron beams just like glass lenses shape light in optical microscopy with Hans Busch inventing the first electromagnetic lens in 1926.  Manfred von Ardenne developed a scanning transmission electron microscope in 1938 that incorporated a tightly focused electron beam shaped with electromagnetic lenses.  In 1942 Vladimir Zwroykin developed the first true SEM with 50nm resolving power.  The first commercial SEM became available in 1965 manufactured by Cambridge Instrument Company, a UK based predecessor of Zeiss Microscopy.

 

Basics

SEM is an extremely versatile technique and often considered the ‘jack of all trades’ due to the fact that depending on the set-up of the instrument it can provide a wide variety of information including surface topography, elemental composition and distribution, crystal orientation, phase distribution and grain size analysis.  In SEM a finely focused beam of electrons is scanned over a sample using a pattern of parallel lines.  A variety of signals (e.g. electrons and x-rays) are emitted due to this interaction and collected to characterize the sample.  SEM provides clear topographical images of sample surfaces with superior resolution (<1nm on some instruments) and depth of field (the distance between the nearest and the furthest objects that are in acceptable focus within an image) than capable with optical microscopy.

​

As the electrons interact with the sample, they lose energy by random scattering and absorption events that occur within a teardrop shaped volume (i.e. the interaction volume).  The ultimate size of the teardrop is dependent on a number of factors including the electron’s landing energy (the actual energy of the electron beam at the sample surface), the atomic number and density of the sample.  

​

As the electron beam interacts with a sample a number of signals are generated that are commonly analyzed in SEM.

 

• Secondary Electrons  (SE):  These electrons are due to inelastic scattering events of the primary electron beam with the atoms of the specimen causing the atom to eject an electron (i.e. secondary electron).  These electrons have very low energies (<50eV) meaning that only those emitted very close to the sample surface (~100nm) will be detected.  Thus, they originate from the top of the teardrop where the radius is smallest resulting in the best topographical resolution of the sample.  Further, by limiting the detection to those secondary electrons formed at the point of interaction with the beam, spatial resolutions as good as 0.4nm have been achieved.

​

• Backscattered Electrons (BSE):  These electrons are those from the primary electron beam after undergoing an elastic scatter off the nucleus of atoms.  Thus, these electrons will have considerably higher energies than secondary electrons allowing them to originate deeper (~500nm) within the sample.  As such, they have worse resolution than secondary electrons.  Further, the intensity of the backscatter event is dependent on the atomic number of the atom.  In other words, backscattering is more likely for larger atoms (higher atomic number) resulting in heavier elements having greater intensity (i.e. appear brighter) than lighter elements in Backscattered Electron images.  Backscattered electrons can also be measured to provide information regarding crystal orientation and phase distribution.  Here the primary electrons enter a sample and diffract along the crystallographic plane.  The diffracted electrons interfere constructively forming a pattern (i.e. the Kikuchi pattern) of intersecting beams representing the reflecting planes at that location in the sample.  The band width and intensity are directly related to atom spacing and the angles between bands are related to the angles between crystallographic planes.  A Hough (mathematical) transform is taken of each Kikuchi band allowing information on orientation to be obtained.

​

• Characteristic X-Rays:  The electron beam can result in the emission of inner shell electrons from atoms.  When an electron from an outer shell drops down and fills the resulting vacancy it causes the emission of an x-ray whose energy is equal to the energy difference between the outer and inner shell.  This energy difference is unique for each type of atom.  Thus, emitted x-rays are characteristic of the types of atoms from which they are emitted and measuring their energies or wavelengths allows sample composition to be determined.

​

• Light (Cathodoluminescence):  The backscattered and secondary electrons can cause electrons in atoms to be excited from their valence band into a conduction band.  As they relax back to their ground state the atoms can emit visible light whose wavelength can be measured allowing the optical activity of a sample to be determined.

 

Samples for SEM analysis must fit inside a vacuum chamber (because electrons are light they are easily stopped by air) and must typically be conductive.  Often, samples are affixed to a metal stub using conductive carbon tape.  Nonconductive samples are often coated with a metal (e.g. Au, Au/Pd, Pt, Ir, W, Cr, graphite) to make them conductive.  This coating is typically done by low vacuum sputter coating or high vacuum evaporation.  Alternatively, specialized SEM instruments (e.g. Variable Pressure SEMS and Environmental SEMs) can be utilized in which the sample sits at relatively high pressure while the electron column is differentially pumped to allow for sufficient vacuum to be maintained within the electron column.  The higher pressures serve to dissipate the charge of nonconductive samples allowing them to be analyzed without coating.  However, these conditions result in worse spatial resolution.

​

Instrumentation

SEM instruments basically consist of an electron source, focusing optics, scanning coils, a sample stage and a variety of detectors all present within a vacuum system maintained via pumps.  The electron source, focusing optics and scanning coils are all located within an electron column sitting on top of the analysis chamber. 

​

Sources

There are three main types of electron sources (i.e. the ‘gun’) used in SEM.

​

• Thermionic:  This source consists of a filament (commonly W wire or LaB6 crystal for higher brightness and longer lifetimes) and works like a light bulb.  The filament is heated to high temperatures (W to 2700K and LaB6 to 1800K) which allows electrons to overcome the work function and be emitted.  Directly below the electron emitter sits a Wehnelt cap in which a small negative voltage (~-200 to -500V) is applied to help shape the electron beam.  Beneath the cap is an anode which is biased positive (typically varies between 0.1-3keV) and is used to accelerate the electrons from the emitter forming the electron beam.  These sources produce very high and stable electron beam currents and generally require less complex vacuum systems.  However, they have shorter lifetimes (several months) and worse spatial resolution (~1.5-4nm) than the other two types of sources.

​

• Cold Field Emitter:  This source uses <310> single crystal W tapered into a sharp tip (can be less than 10nm radius).  No heat is used.  Rather, high electric field (~3-4keV) is applied to the tip via a metal grid (i.e. the First Anode) allowing electrons to be stripped from the tip via tunneling.  A second anode sitting below the first is used to set the acceleration voltage like with a thermionic emitter.  The Cold Field Emitter source can achieve high resolution (0.4-0.6nm) and has longer lifetimes (several years) than the Thermionic source.  However, it has lower probe current and stability.  It also requires better vacuum within the gun and needs periodic flashing (i.e. heating) to clean the tip since it operates cold allowing contaminants to adsorb onto the tip surface over time.

​

• Thermal Field Emitter:  This emitter is similar to the Cold Field Emitter but uses a heated (1800K) W <100> single crystal coated with a ZrO2 film to lower the surface work function.  This source has higher probe currents, better current stability and doesn’t require as good of a vacuum as the Cold Emitter but its resolution (0.5-0.8nm) and lifetime (~2yrs) aren’t quite as good.

​

Lenses

In an SEM the lenses used to shape and focus the electron beam consist of coiled Cu wires wrapped around a soft iron cylinder (i.e. pole piece).  Current is passed through the coils which produces a magnetic field that modifies electron trajectories through the lens via the Lorentz Force (F).

              

                       F=q[E+(v x B)]     where

                      

                       q:  Electron charge

                       E:  Electric Field              
                       V:  Velocity
                       B:  Magnetic Field 

​

The magnetic field strength is controlled by adjusting the current through the Cu coil.

​

Typically, an electron column contains 1-3 lenses with 3 being most common.  There are two types of lenses.  The first lenses in the column are Condenser lenses (will be 1 or 2) which form the electron beam into a smaller diameter.  The final lens is commonly called the Objective lens which focuses the beam onto the sample with the smallest diameter possible.   The user adjusts the current in this lens during the focusing process.  Most SEM instruments use a ‘Conventional’ objective lens in which the magnetic field is present within the lens and the sample sits below the lens allowing the sample to be freely moved within the instrument.  However, because the distance between the lens and the sample can be large, a long focal length is required resulting in larger aberration which limits resolution.  Thus, for high resolution instruments a Semi-In-Lens (Snorkel) Objective Lens is sometimes used where the polepiece shape of the lens is designed to produce the magnetic field below the lens allowing for shorter focal lengths between the field and the sample and ultimately higher resolution.

​

Also present within the column are a number of apertures that are used to prevent scattered electrons from traveling down the column (fixed apertures) and to define the convergence angle (Movable Apertures).  The convergence angle describes the angular range over which electrons are incident at each point on the sample and ultimately controls the resolution, depth of field and beam current.  Typically, a Movable Aperture that results in higher beam currents has a larger beam spot size and poorer depth of field and vice versa.

​

The electron beam is rastered across the sample surface using deflector plates consisting of two pairs of scanning coils located in between the 2nd Condenser Lens and the Objective Lens.  The signal produced at each location on the sample is collected allowing detailed imaging and elemental analysis to be obtained.

​

The analysis chamber contains a sample stage and a number of detectors capable of analyzing the various signals produced from the interaction of the electron beam with the sample.  Sample stages typically provide x-y-x translation as well as tilt and rotation so that the sample can be manipulated to any position necessary for the analysis being performed.

​

Detectors

The most common detector in SEM is the Everhart-Thornley secondary electron detector.  A slightly positively biased grid (i.e. the Faraday cage) is used to attract the low energy (<50eV) secondary electrons to the detector where they hit a scintillator (e.g. phosphor) held at high positive potential (~10keV) and emit light (i.e. the electron signal is converted to a light signal).  The light travels down a light pipe before striking a photomultiplier producing photoelectrons.  The photoelectrons are accelerated towards a series of dynodes held at positive bias (~100eV) and containing a high yield coating which gives off additional electrons resulting in amplification of the initial electron signal.  Total effective amplification of the PM/Scintillator combination is typically 108 resulting in a high signal to noise ratio and allowing the detector to produce high-resolution stereoscopic images of the sample surface.

​

The secondary electron detector typically sits to the side of the sample stage.  However, these detectors are also more commonly being found on instruments within the electron column immediately above the objective lens.  These are called ‘in lens’ or ‘through lens’ detectors.  These detectors are better at limiting secondary electron detection to those that are only generated from the primary beam’s impact point (SE1 electrons) and excluding those from secondary electrons generated during the interaction of  multiple primary beam scatter events inside the interaction volume which can be emitted from the sample at a greater distance from the interaction point (SE2 electrons).  The in lens detectors also avoid detecting secondary electrons generated from backscattered electrons colliding with the chamber walls and electron column (SE3 electrons).  This discrimination results in better resolution and more surface sensitivity revealing the presence of surface films and contaminants that are often invisible using conventional Everhart Thornley detectors.

​

The backscattered electron detector is most commonly a solid-state device (i.e. semiconductor based) and is situated between the sample and the objective lens.  A hole exists in the detector allowing the primary beam to pass through it as the beam travels to the sample.  The backscattered electrons are then collected within the detector as they return back towards the electron column after scattering from the sample.  Some BSE detectors consist of a number of quadrants in which the signal can be collected independently from different directions allowing multiple images to be obtained simultaneously and subsequent 3D reconstruction possible with appropriate software manipulation.  Backscattered electron yield is dependent on a number of factors including atomic number (Z) of the sample, the specimen angle with relation to the primary electron beam and the primary beam voltage.  Thus, the backscattered electron images provide information based both on the sample composition and surface topography.  In backscattered images material with high Z will appear brighter (i.e. they have a higher yield) while lower Z material will appear darker due to the lower probability of backscatter occurring from them.

​

X-ray detection is accomplished with an Energy Dispersive Spectroscopy (EDS) or a Wavelength Dispersive Spectroscopy (WDS) detector with the former being most frequently used.  Most commonly the EDS detector consists of a Si(Li) crystal operating at low voltage that absorbs the incoming x-ray energy via ionization.  This ionization produces free electrons within the crystal that become conductive creating an electrical charge bias that is proportional to the x-ray energies which are characteristic of the elements from which they were generated.  The detector is cooled with liquid nitrogen and is mounted within the instrument at the end of a long arm protruding from the instrument with a liquid nitrogen dewar attached.  Because the charge signals produced by the detector are small, cooling is necessary to reduce the electronic noise of the detector and prevent it from interfering with the x-ray signals.  However, some EDS detectors use newer technology utilizing silicon drift detectors which have higher count rates and preclude the need for liquid nitrogen cooling.

​

EDS spectra consist of a plot of Intensity (y-axis) versus x-ray energy (x-axis) with the signals corresponding to the characteristic x-rays generated from the sample under study.  While the signal intensity is proportional to the amount of a particular element present a number of factors can affect x-ray production meaning that without a homogeneous sample and good standards quantification should be considered roughly semi-quantitative, at best.

​

Initial EDS analysis typically involves generating an X-ray spectrum from the entire SEM scan area.  However, since signal is produced from each location of the rastering electron beam, elemental maps can be created showing the distribution of the various elements within a sample.  While SEM images can have sub-nm spatial resolution, that of elemental maps is in the 1-3um range due to the large interaction volume in which x-rays are produced when an electron strikes a particular spot on the sample.

​

While EDS detectors are very good at performing quick, full elemental analyses, energy overlaps from different elements can make it difficult to distinguish between some elements.  Detection limits are typically in the range of 0.1-0.5wt% with it being considerably worse (a few atomic %) for lighter elements.  Additionally, EDS can’t detect lighter elements (below atomic number of Na) for detectors typically equipped with a Be window.

​

The other method of detecting the emitted X-rays is via WDS.  As the name indicates, while EDS measures the energies of the emitted X-rays formed upon interaction of the electron beam with the sample, WDS instead measures the wavelengths of the X-rays.  Both energies and wavelengths are characteristic of the specific atom from which the signal is generated.

​

WDS uses crystals to separate the x-rays emitted by the samples into individual wavelengths.  The x-rays pass into the WDS system through a window (typically Be) and onto a crystal which diffracts a single wavelength onto the detector measuring the relative quantities of each wavelength that arrives.  Whether an x-ray is diffracted by the crystal onto the detector will depend on its wavelength, the orientation of the crystal and the crystal lattice spacing.  The wavelength diffracted onto the detector can be changed by modifying the crystal orientation.  However, even by changing the orientation the crystal will be unable to diffract all possible wavelengths to the detector that may be present.  Thus, multiple crystals with dissimilar lattice constants are used to access the entire range of wavelengths of interest.  Crystals can be minerals, metallic, organic and synthetic multilayers.

​

WDS has much better resolution (~5-20eV) than EDS (typically ~130-140eV) resulting in much less signal overlap/interference.  Its detection limits (~0.01wt% for heavier elements) are also better than EDS due in part to its lower background noise.  Additionally, quantification is better in WDS as it has significantly higher reproducibility than EDS.  However, WDS systems are typically slower than EDS because of the need to scan through the various wavelengths of interest and WDS systems are relatively more expensive. 

​

Another detector that is becoming more commonly found on SEM instruments is the Electron Backscattered Diffraction detector (EBSD) which is used to measure local crystal orientation, phase distribution (note it’s not good for identifying individual phases) and grain size analysis, all with submicron spatial resolution.  When analyzing a sample by EBSD the sample is tilted to 70 degrees allowing more electrons to be backscattered from the surface.  When the electron beam hits the sample the backscattered electrons are collected onto a phosphor that transforms the electrons into photons.  The diffracted portion of the backscattered electrons is extremely anisotropic (i.e. directionally dependent) such that when they interact with the phosphor they create a pattern called a Kikuchi pattern.  These are formed by the scattering of electrons from each individual plane on the sample surface.  Thus, different patterns are obtained depending on the orientation of the sample and the types of crystals present.  The Kikuchi pattern is captured by a CCD camera behind the phosphor screen which is then transferred to the computer.  These patterns can be collected from each location of the electron beam on the sample allowing mapping to determine where orientations and grain boundaries exist.

​

In EBSD amorphous materials produce no pattern.  Thus, ideal samples for this measurement are materials such as metals, alloys and minerals.  EBSD is often used to characterize annealing techniques, understand crystallographic growth mechanisms and as a failure analysis tool in metallurgical applications.

​

Aberrations

In SEM, the ultimate resolution is determined by both the spot size of the beam and the interaction volume (which has been shown above to be larger than the beam diameter).  A number of limiting parameters exist within an SEM to reduce the resolution. 

​

• Spherical Aberration:  The magnetic field within the Condenser/Objective lenses are not uniform being stronger towards the outside and weaker towards the center.  Thus, electrons that are off-axis will be over focused.  A limiting aperture is often present to remove the over focused electrons. 

​

• Chromatic Aberration:  Electrons with different energies will be focused to different points along the image plane.  Sources that are heated (e.g. Thermionic, Schottky) result in more energy spread compared to those that aren’t (e.g. cold field emission).  The energy spread can be decreased by increasing the Acceleration Voltage (i.e. the electron energy). 

​

• Aperture Diffraction:   Because electrons follow the same physics as light waves they can also diffract (i.e. bend) as they pass through a small aperture. 

​

• Astigmatism:  Small defects in the manufacturing of the lens (e.g. the pole pieces and Cu coils) can result in stray magnetic fields that can affect the electron beam.  A set of deflectors that are perpendicular to the optical axis are used to account for this issue by shaping the beam to bring it into focus.  

         

Comparison

SEM is used in many different applications due to its ability to provide a wide variety of easy to interpret information quickly and with relatively easy sample prep.  While it has excellent resolution over a wide range of magnification, AFM and TEM can provide better resolution.  While it can provide elemental composition using EDS or WDS, it’s typically not considered accurate, in particular without standards.  For more accurate quantification and better detection limits WD-XRF should be considered.  Additionally, SEM elemental analysis is considered a bulk technique.  If surface information is needed AES and/or XPS are better options.  While EBSD can provide surface orientation and texture analysis, it can’t provide detailed phase information.  Rather, XRD needs to be used for that analysis.