465

Scanning electron microscopy for materials characterization

David C Joy

Current materials are usually complex in chemistry, for heating, cooling, tensile or other special devices and
three-dimensional in form, and of rapidly diminishing because such experiments provide uniquely valuable data.
microstructural scale. To characterize such materials the
scanning electron microscope (SEM) now uses a wide range Driven by the needs of the semiconductor industry,
of operating conditions to target the desired sample volume, however, the predominant use of the SEM is as an
sophisticated modeling techniques to interpret the data. analytical imaging tool for the metrology of integrated
It also uses novel imaging modes to derive new types of circuits. Here the requirement is to be able to extract from
information. These include depth-resolved three-dimensional an image precise and accurate measurements of selected
data, and spatially resolved crystallographic data. features for the control of the fabrication processes. This
has been made feasible as a result of the development
of detailed Monte Carlo models of the electron solid
Addresses interaction occurring in the SEM [6**,7,8”]. It is now
Electron Microscopy Facility, Department of Biochemistry,
routinely possible to take an experimental image and then
Cellular and Molecular Biology, University of Tennesse, Knoxville,
TN 37996-0810 and High Temperature Materials Laboratory,
to compare it with a computed simulation of the same
Oak Ridge National Laboratory, Oak Ridge, TN 37931, USA; image, and by refining this match to iteratively deduce the
e-mail: djoy@utk.edu parameters which describe the three-dimensional structure
Current Opinion in Solid State & Materials Science 1997, of the surface [9,10*]. Although difficult to apply to
2465-466 poorly characterized situations [ 1l] the technique is quite
successful when applied to the well defined geometry
Electronic identifier: 1359-0266-002-00465
encountered in semiconductor devices, although problems
0 Current Chemistry Ltd ISSN 1359-0286 arising from sample charging are still a limit to the accuracy
Abbreviation that is possible.
SEM scanning electron microscope
While the SEM is usually considered to primarily be a tool
for the visualization of surfaces, the electron beam does
penetrate many micrometers into the sample. Because
Introduction
the penetration increases rapidly with the incident energy
The scanning electron microscope (SEM) is the most
this effect can be exploited to add the missing depth
versatile and widely used electron beam instrument in
dimension to imaging. For example, by choosing the
the world. It owes its popularity to the easily interpreted
energy of the incident beam and of the collected
nature of the micrographs that it generates, to the
backscattered electrons, information specific to a particular
diversity of types of information that it can produce,
depth beneath the surface of a specimen can be retrieved
and to the fact that images and analytical information
[12*]. A sequence of images obtained by appropriately
can readily be combined. As this review demonstrates
varying the incident and exit energies then produces a
the use of the SEM for materials characterization is
sequence of optical sections of the sample which can
increasingly motivated by the desire to obtain not just
be reconstructed into a full three-dimensional image.
images but quantitative information in two, or even
Because of the multiple scattering events experienced
three dimensions, about the microstructure, the chemistry,
by every electron [13,14”] there is significant mixing of
the crystallography and the electronic properties of the
information between successive layers and so the vertical
material of interest.
resolution is limited, but the method is a further important
step in moving the SEM image from two into three
SEM for materials characterization dimensions and provides a non-destructive visualization
The basic mode of use of the SEM has always been in the tool of significantly higher lateral spatial resolution than
imaging of surface topography, and 90% of the literature is X-ray microtomography.
still devoted to one aspect or another of this application.
The increasingly high level of the performance of the In some parallel developments it has been shown that a
SEM, in particular, makes it an excellent complement to sequence of cathode-luminescence (emitted light) images
high resolution transmission electron microscope imaging produced from suitable samples, as the beam energy
[1*,2] (as it provides the surface detail lacking in the is varied, can also be processed to yield a stack of
TEM micrograph) or to scanning tunneling or atomic force depth resolved slices through the specimen [1.5*,16]. This
microscopies, where it can provide images for comparison cathode-luminescent (CL) signal is produced by the decay
that are readily interpretable [3*]. The SEM is increasingly of the electron-hole pairs generated by the incident beam
being used for in situ and dynamic experiments [4,5] in a semiconductor or insulating material and the intensity
because there is ample room in the specimen chamber of the CL signal, as well as the form of its spectrum,
466 Characterization techniques

depends on the type and concentration of dopant species The use of the SEM at low beam energies, that is to
in the material. So in this case the image is a map of say below 5 keV, has now become the preferred mode
the volume distribution of the chemistry as well as of the of operation for many applications [28]. The attraction of
external and internal morphology of the material. the technique is that it permits the successful observation
of materials which are difficult or impossible to image
The electron-hole pair signal can also be detected from the satisfactorily at higher beam energies. This is, in part,
sample by extracting it as a current-the ‘electron beam because the increase in secondary electron emission at
induced current’ (EBIC) signal-under the influence of a low energies can result in a state of dynamic charge
field from either internal P-N junctions or from a surface balance at the specimen so that poorly, or nonconducting,
Schottky barrier. The EBIC signal provides a unique specimens can be imaged without net charge build-up
internal view of a semiconducting material, permitting [29**]. It is therefore not necessary to use any surface
direct visualization of all active junctions as well as of coating to confer conductivity allowing the true surface
quantum wells, and electrically active defects in the structure of the sample to be observed. At the same time
material [17]. Once again if a sequence of images is the reduction in beam energy results in a decline in the
obtained as a function of incident beam energy then depth depth of penetration of the electrons so that radiation
resolved images can be generated [18*] which in this case induced damage is localized at the surface rather than
represent a volume map of parameters such as the minority spread through a large volume. The LVSEM (low voltage
carrier diffusion length in the specimen. SEhsI) is therefore well suited for the study of polymers
[30-323 and other fragile and exotic materials such as snow
A significant strength of transmission electron microscopy crystals [33”].
(TEM) as a materials characterization technique has been
the availability of various modes of electron diffraction for
the determination of crystal structure and for the imaging Microanalysis is an essential component of materials
of dislocations, grain boundaries, and other lattice defects, characterization and has been an important application
crucial to the behavior of the material. The development of the SEM since energy dispersive X-ray spectrometers
of the technique of electron channeling [ 191 and the arrival first became available thirty years ago. However, the
of high brightness field emission guns on SEMs has now demands placed on microanalysis in the SEM are also
made it possible to obtain both crystallographic orientation rapidly changing. Historically, microanalysis was based
information from sub-micrometer areas, such as is required on the assumptions that the specimen to be analyzed
for grain boundary studies [ZO]; and also to directly image was chemically homogeneous over at least the beam
dislocations in a bulk specimen [Zl]. An entirely different interaction volume of a few micrometers diameter, that
approach to the determination of crystallography in the the sample was flat, and that the surface was smooth.
SEhl is the technique of electron backscattering patterns For the problems encountered in today’s materials science
(EBSP). In this mode of operation the angular distribution none of these assumptions are likely to be valid and
of the backscattered electrons is imaged by a suitable so new procedures for obtaining and interpreting data
detector and then analyzed automatically by a computer to are required. With the continual reduction in the scale
yield the hIiller indices representing the crystallographic of microstructure the spatial resolution of an analysis is
orientation. A spatial resolution of the order of 1OOnm now a crucial issue. While the fundamental limits set by
is possible [ZZ’] with an angular precision of better than the electron-solid interactions remain, the application of
one degree [23.*]. In fact, with careful experimental hlonte Carlo simulations demonstrate that small features,
control angular rotations as small as 0.01 degrees can be such as a monolayer of an element at a grain boundary,
detected, which corresponds to an elastic strain within can reliably be detected under optimized conditions [34**].
the sample of only about 0.02% [24”], a value typical An alternative route for improving spatial resolution is
of the magnitude of the strains found in high quality the use of low beam energies which offer much reduced
epitaxial films. A unique feature of this technique is the beam interaction volumes but this raises the practical
ability to synthesize an ‘Orientation Image’ in which, for problem that only low energy X-ray lines are accessible
example, the three colors of an RGB (red-green-blue) and so peak overlaps become a major limitation. However,
color-coded display are employed to represent the spatial new bolometric device technology for X-ray spectrometers
variation of the three Euler angles describing the crystal [35**] offer sufficiently high energy resolution such chat
orientation. These orientation images reveal clearly, and all probable peak overlaps can be separated. Some novel
display quantitatively, effects such as texturing [25’], strategies for microanalysis are also under development.
recovery and recrystallization [26] and grain boundary For example, the secondary electron image can reveal, in
misorientations [27*]. Although the time taken to derive a quantitative manner, the concentration of dopants in a
such an image is often many hours the detail and quality semiconductor [36”] even at levels corresponding to only
of the information is unmatched by any other technique about one part per million, far below the concentrations
and it can be claimed fairly that the crystallographic required for X-ray analysis. Secondary electron imaging
capabilities of the SEhI now exceed those of the TEhl used in this way is, in fact, closely related to Auger analysis
in their versatility and power. because it is a measure of the total electron yield [37’]
 Scanning electron microscopy for materials characterization Joy 467

and if it can made fully quantitative, then it may be a 3. Brostin J: Scanning probe microscopy of polymers in the field-
. emission SEM. In Microscopy and Microanalysis 7996. Edited by
promising new tool for characterization. Bailey GW, Corbett JM, Dimlich R, Michael JR, Zaluzec NJ. San
Francisco: San Francisco Press Inc; 1996:676-678.
Excellent example of the synergism between the SEM and scanned probe
The single most exciting recent contribution of the
microscopies.
SEM to materials characterization has been through the
4. Melikhov IV, Nikolaev AL, Bozhevolnov VE, lvanov LN,
development of the environmental (ESEM) or nature Kozlovskaya ED, Saparin GV: A study of the kinetics of
SEM (NSEM). In these instruments the vacuum condition topochemical reactions by SEM. SCANNING 1995,17:371-
376.
around the specimen has been relaxed so that the
5. Dorozhkin SV: Procsss of epitaxial crystal growth for C&O, on
conventional high vacuum has been replaced by a gaseous
a surface of dissolving fluorapatfte crystals studied by SEM.
environment in which the pressure can be as high as 20 SCANNlNG 1996, 15119-l 24.
torr (2,300Pa). The presence of a gas, typically saturated 6. Lowney JR: Monte Carlo simulation of scanning electron
water vapor or moist air, has significant effects on both .. microscope signals for lithographic metrology. SCANNfNG
1996, l&291-301.
the imaging mechanisms and on the types of sample that A definitive statement of the computational and practical techniques required
can be examined. Ion production in the gas results in for the use of the SEM as a metrological tool.
dynamic charge neutralization of the specimen permitting 7. Joy DC: Monte Carlo Simulation for Microscopy and
poorly conducting materials to be successfully imaged at Microanalysis. New York: Oxford University Press; 1995.

high beam energies (38*,39’,40) as well as producing novel 6. Ding Z-L, Shimiru R: Monte Carlo modeling of electron
interaction with solids. SCANNlNG 1996, 15:92-l 14.
imaging effects such as the ability to see to the bottom of gutstanding mathematical description of the processes involved
deep cracks and holes [41”]. Most importantly the ESEM in electron-solid interactions.
and NSEM configurations liberate the user from the prob- 9. Sartore RG: Using secondary electrons to obtain improved
lems associated with preparing specimens for observation linewidth measurements. In Microscopy and Microanalysis
7996. Edited by Bailey GW, Corbett JM, Dimlich R, Michael JR,
in conventional high vacuum SEMs. An enormous range Zaluzec NJ. San Francisco: San Francisco Press Inc; 1996:944-
of real-world problems immediately become accessible for 946.

examination in this way for example, the in situ observation 10. Howell PGT: A computer program to illustrate macro-
. topography on electron backscattering. SCANNlNG 1996,
of corrosion [42’], the formation of cement [43*], the 15426433.
study of pharmaceuticals [44] and common household and Good review of the procedures required to interpret surface topography in
a quantitative manner by employing Monte Carlo simulations.
personal products [45’]. As most of the standard imaging
and analysis modes of the SEM can also be performed 11. Czepkowski T, Slowko W: Some limitations of surface profile
reconstruction in SEM. SCANNlNG 1996, 16433-447.
in the gaseous microscope, the potential scope of this
12. Rau El, Robinson VNE: An annular toroidal backscattered
machine for real materials characterization is considerable. . electron energy analyser for use in scanning electron
The interaction of the electron beam with the gas can microscopy. SCANNING 1996, 16:556-562.
even be used to deposit metal on the specimen surface The combination of a selected incident energy and a selected imaging en-
ergy provides depth-resolved imaging.
for the fabrication of connector lines [46”] or to form
13. Gerard P, Balladore JL, Martinez JP, Ouabou A: Experimental
micro-miniature three-dimensional structures [47”]. The determination of angular-energy distributions of electrons
SEM as a tool for materials characterization thus becomes backscattered by bulk gold and silicon samples. SCANNlNG
1995, 17:377-307.
in turn a new tool for the fabrication of still more complex
materials. 14. Gauvin R, Drouin D, Hovington P: Energy-filtered backscattered
.. images of 10 nm NbC and AIN precipitates in steels computed
by Monte Carlo simulations. In Microscopy and Microanalysis
Condusions 7996. Edited by Bailey GW, Corbett JM, Dimlich R, Michael JR,
Zatuzec NJ. San Francisco: San Francisco Press Inc; 1996:146-
While the basic capabilities of the SEM have remained 150.
unchanged, the combination of innovative operating Detailed calculations which support the principle of depth-resolved imaging
discussed in reference 1131.
strategies and the use of Monte Carlo, or other, modeling
techniques has resulted in a significant enhancement in 15. Saparin GV, Obyden SK, lvannikov PV: A nondestructive method
. for three-dimensional reconstruction of luminescent materials;
the variety and quality of the data that can be obtained principles, data acquisition, image processing. SCANNlNG
from materials. Thus, while the complexity of the task 1996, 15261-291.
Innovative approach to three-dimensional reconstruction of both chemical
of microcharacterization continues to increase the SEM and morphological data
remains fully capable of meeting these challenges. 16. Saparin GV, Obyden SK: Morphologic and cathodolumines-
cence studies of diamond films by scanning electron mi-
croscopy. SCANNING 1995, 17:337-340.
References and recommended reading
Papers of particular interest, published within the annual period of review, 17 Drouin D, Gauvin R, Beauvais J, Hovington P, Joy DC:
Characterization of Schottky depletion zone using EBIC
have been highlighted as:
imaging. In Microscopy and Microanalysis 1996. Edited by
. Bailey GW, Corbett JM, Dimlich R, Michael JR, Zaluzec NJ. San
of special interest
l * of outstanding interest Francisco: San Francisco Press Inc; 1996:150-l 52.

16. Hardingham C: Use of electron-beam induced current in a SEM

1. Ajayan PM, Redlich P, Ruble M: Structure of carbon nanotube-
based nanocomposites. J Microsc 1997,165:275-262.
;he high resolution surface images of an SEM provide complementary data
to structural infonnation obtained from a TEM.
2. Yang W, Weathetfy GC, McComb DW, Lloyd DJ: The structure of
Sic-reinforced Mg casting alloys. J Microsc 1997, 155:292-302.
for analysis of space solar cells. J Mater Sci 1996, 31:661-666.
Lcellent demonstration of the power of the EBIC technique in the SEM for
the characterization of complex three-dimensional device structures.
19. Joy DC, Newbury DE, Davidson DL: Electron channeling patterns
in the scanning electron microscope. J Appf Phys 1962,
53:RBl -R102.
468 Characterization techniques
20. Watanabe T: SEM-ECP analysis of grain-boundary character
distribution in oolvcrvstalline materials. In Microscoov and
Microanalysis 1’996. idited by Bailey GW, Corbett Ji,‘Dimlich R,
Michael JR, Zaluzec NJ. San Francisco: San Francisco Press Inc;
1996:354-356.
21. Simkin BA, Crimp MA: Dislocation detection depth
measurements in silicon using electron channeling contrast
In Microscopy and Microanalysis 7996. Edited by Bailey GW,
Corbett JM, Dimlich R, Michael JR, Zaluzec NJ. San Francisco: San
Francisco Press Inc; 1996:140-l 42.
22. Kenik EA: Spatial resolution of EBSP diffraction in a FEG-SEM.
. In Microscopy and Microanalysis 1996. Edited by Bailey GW,
Corbett JM, Dimlich R. Michael JR, Zaluzec NJ. San Francisco: San
Francisco Press Inc; 1’996:348-350.
Important experimental determination of the actual spatial resolution of EBSP
analysis.
23. Lassen NCK: The relative precision of crystal orientations
measured from EBSP. I Microsc 1996, 181:72-61.
rareful analysis of the errors which set a limit to orientation analysis using
EBSP patterns.
24. Wilkinson AJ: Measurement of elastic strains and small lattice
..
rotations using EBSP. Ultramicrosc 1996, 62:237-242.
Demonstrates that lattice rotations as small as those associated with elastic
deformation and resiudal strain can be observed and measured using the
EBSP technique.
25. Sutcliff JA. Bewlav BP: Orientation imaaina of a Nb-Ti-Si
. directionally solidified in-situ composite. in Microscopy and
Microanalysis 7996. Edited by Bailey GW, Corbett JM, Dimlich R,
Michael JR, Zaluzec NJ. San Francisco: San Francisco Press Inc;
1996:350-351.
Good illustration of the power of the orientation imaging technique for ma-
terials characterization.
26. Wright SI, Dingley DJ, Field DP: Determining deformation,
recovery and recrystallization from orientation imaging
microscopy. In Microscopy and Microanalysis 1996. Edited by
Bailey GW, Corbett JM, Dimlich R, Michael JR, Zaluzec NJ. San
Francisco: San Francisco Press Inc; 1996:352-354.
27. Caul MV, Randle V: Grain boundary characterization in
. austenitic steel. In Microscopy and Microanalysis 7996. Edited
by Bailey GW, Corbett JM, Dimlich R, Michael JR, Zaluzec NJ. San
Francisco: San Francisco Press Inc; 1996:344-345.
Quantitative analysis of grain boundary quality and character using EBSP
techniques.
28. Joy DC, Joy CS: Low voltage scanning electron microscopy.
Micron 1996, 27:247-265.
29. Cazaux J: Electron probe microanalysis of insulating materials.
.. X-ray Spectrometry 1996, 27:265-280.
A definitive analysis of the problems of using electron beams to study non-
conducting, or poorly conducting, materials.
30. Brown GM, Butler JH: A new method for the characterization of
polymer blends using LVSEM. In Microscopy and Microanalysis
7996. Edited by Bailey GW, Corbett JM, Dimlich R, Michael JR,
Zaluzec NJ. San Francisco: San Francisco Press Inc; 1996:l QO-
192.
31. Butler JH, Joy DC, Bradley GF, Krause SJ: Low voltage scanning
electron microscopy of polymers. Polymer 1995, 36:1781-l 790.
32. Festag R, Wunderlich B, Joy DC, Alexandratos SD, Cook K:
Separation and characterization of single-chain polymer
particles. Proc W/E 1996, 2809:155-l 65.
33. Wergin WP, Rango A, Erbe EF: Use of low temperature SEM
.. to observe icicles, ice fabric, rime, and frost In Microscopy and
Microanalysis 7996. Edited by Bailey GW, Corbett JM, Dimlich R,
Michael JR, Zaluzec NJ. San Francisco: San Francisco Press Inc;
1996:146-l 47.
Beautiful application of the SEM to the study of the fragile and complex
microstructure of snow, ice and frost. A technical tour de force.
34. Wilson AR: Monte Carlo and finite element calculations of
.. X-ray production at interfaces in bulk materials. Ultramicrosc
1996, 66:117-l 31.
A demonstration that the spatial resolution limit to microanalysis set by
electron-solid interactions does not prevent the observation of small features
provided that conditions are properly optimized.
35. Silver E, Legros M, Madden N, Beeman J, Hailer E: High
.. resolution, broad band microcalorimeters for X-ray
microanalysis. X-ray Spectrometry 1996, 26:l 15-l 22.
The development of a new type of energy dispersive X-ray detector offering
an order of magnitude better performance than current detectors.
36. Turan R, Perovic DD, Houghton DC: Mapping electrically active
.. dopant profiles by field emission SEM. Appl Pbys Left 1996,
69:1593-l 595.
Secondary electron signals can reveal the presence of dopants at levels
consistent with those used in device fabrication. An important new analytical
tool in the SEM.
37. Schroeder S L M: Towards a ‘universal curve’ for total electron-
yield XAS. Solid State Commun 1996, 98:405-409.
inalysis of the behavior of the emitted secondary electron signal as it is af-
fected by Auger emissions. In conjunction with reference [36’*] this provides
a prototype new technique for microanalysis in the SEM.
38. Meredith P, Donald AM: Study of ‘wet’ polymer latex in the
. environmental SEM. J Microsc 1996, 181:23-36.
Demonstration of the power of the ESEM to study real world problems.
39. Meredith P, Donald AM, Thield B: Electron-gas interactions in
the ESEM gaseous detector. SCANNlNG 1996, l&467-473.
&ood analysis of the complex interactions encountered by electrons in the
gas of the ESEM.
40. Joy DC: Modeling the electron-gas interaction in low-
vacuum SEM. In Microscopy and Microanalysis 7996. Edited by
Bailey GW, Corbett JM, Dimlich R, Michael JR, Zaluzec NJ. San
Francisco: San Francisco Press Inc; 1996:836-837.
41. Newbury DE: Imaging deep holes in structures in the ESEM.
SCANNING 1996 18~474-483.
%e conversion of second& electrons to ions in the gas of the ESEM spec-
imen chamber makes it possible to form satisfactory images from the bottom
of holes several millimeters in depth.
42. Pirttiaho L, Blakely J: ESEM observations of H,S attack on the
. protective oxide of an Ni-Fe alloy. SCANNlNG 1996. 18:497-
500.
Good real world application of ESEM technology.
43. Neubauer CM, Jennings HM: The role of the ESEM in the
. investigation of cement-based materials. SCANNlNG 1996,
18:515-521.
Application of the ESEM to the classic problem of the behavior of cement
and related materials.
44. D’Emmanuele A, Gilpin C: Applications of the ESEM to the
analysis of pharmaceutical formulations. SCANNING 1996,
18:522-528.
45. Hoyberg K: ESEM of personal and household products.
. SCANNlNG 1996. 19:109-l 13.
Interesting demonstration df the versatility of the ESEM for the characteriza-
tion of consumer products.
46. Folch A, Servat J, Esteve J, Tejada J, Seco M: High vacuum
.. versus environmental electron beam deposition. J Vat Sci
Techno/ 1996, 14:2609-2614.
Organometallic gases deposit metal films on to a substrate under electron
beam impact in the SEM. A new tool for lithography and localized chemistry.
47. Paul BK, Klimkiewicz M: Application of ESEM to micro-
.. mechanical fabrication. SCANNlNG 1996, 18:490-496.
The use of the ESEM to build three-dimensional nano-scaled structures us-
ing selective deposition and radiation techniques.