instrument. FIB-SEM as we know it today (sometimes called FIB-SEM tomography), started to be used in materials science at the end of the 1990s [25, 26]. Since
this time, there are continual improvements to this powerful tool as it spreads into
all aspects of biological research for imaging in 3D at nanoscale resolutions [6].
There are several additional factors to be considered for the serial block-face
techniques that are not an issue when performing standard ultrathin section TEM.
Importantly, the interaction between the electron beam and the sample leads to an
accumulation of charges at the sample surface, which are deleterious to the imaging
quality. While high voltages allow higher resolutions, charge formation is normally
inversely proportional to the acceleration voltage. Hence, the balance between
image quality and sample damage is critical. SEM has always suffered from
charging problems since samples are extremely susceptible to damage from the
electron beam and have to be imaged at low voltages. A factor that is rarely
considered in normal TEM work where ultrathin sections with little mass are held
on a highly conductive metal grid. Coating the entire surface of a sample with metal
alleviates the issue for normal SEM, however having a sample embedded in a
non-conductive resin with its surface fully exposed to the electron beam becomes
an issue of real concern to the serial block-face techniques, especially SBEM [27,
28]. The beam damage can take several forms, such as charging, image distortions
forming lines or shifting images, uneven sectioning thickness, or poor sectioning
resulting in a damaged surface. In SBEM, charging is the most limiting factor to the
technique. In FIB-SEM, whether imaging or milling, the ion beam introduces Ga
þ
ions on/into the sample as well as destroying its surface [29]. These positive gallium
ions reduce the charging affect of biological samples, having a similar effect as gas
in variable pressure-SEM, to neutralise the charge at the sample surface [29, 30].
The SBEM, while it does not give as good of a lateral and axial resolution as
FIB-SEM, still has many advantages. The removal of material by a diamond knife
means the process is very fast, taking only a few seconds to remove a section from
the entire block-face. Nevertheless, this is a limiting factor, since the size of the
sample is restricted to the width of the diamond knife (typically 1.2–1.5 mm) and
should be as thin as possible to reduce the charging volume. The FIB only allows
milling of a relatively small area (in the range of 50 Â 50 µm) at any one time and
the area of this exposed face dictates the speed at which the cutting can take place.
Thus, this technique is well suited to smaller volumes and achieving high resolutions. Titze and Genoud [10] give a comparison of the best voxel sizes achieved
(x, y, z): 10 Â 10 Â 25 nm in SBEM, [31] and 5 Â 5 Â 5 nm for FIB-SEM
[32, 33]. However, Russell et al. [9] have shown that they could produce a large
dataset cutting slices at 10 nm and thus giving isotropic resolution for the SBEM of
10 Â 10 Â 10 nm.
The interplay between resolution and size is a problem common to all imaging
techniques [34]. The best image quality from the region of interest (ROI) would be
achieved by a small pixel size, high dwell time (the amount of time that the beam
will collect data at each pixel) and often a large FOV. However, this is rarely
achievable at the same time. The stability of the sample and the system in use will
5 Volume Scanning Electron Microscopy: Serial Block-Face …
121
this time, there are continual improvements to this powerful tool as it spreads into
all aspects of biological research for imaging in 3D at nanoscale resolutions [6].
There are several additional factors to be considered for the serial block-face
techniques that are not an issue when performing standard ultrathin section TEM.
Importantly, the interaction between the electron beam and the sample leads to an
accumulation of charges at the sample surface, which are deleterious to the imaging
quality. While high voltages allow higher resolutions, charge formation is normally
inversely proportional to the acceleration voltage. Hence, the balance between
image quality and sample damage is critical. SEM has always suffered from
charging problems since samples are extremely susceptible to damage from the
electron beam and have to be imaged at low voltages. A factor that is rarely
considered in normal TEM work where ultrathin sections with little mass are held
on a highly conductive metal grid. Coating the entire surface of a sample with metal
alleviates the issue for normal SEM, however having a sample embedded in a
non-conductive resin with its surface fully exposed to the electron beam becomes
an issue of real concern to the serial block-face techniques, especially SBEM [27,
28]. The beam damage can take several forms, such as charging, image distortions
forming lines or shifting images, uneven sectioning thickness, or poor sectioning
resulting in a damaged surface. In SBEM, charging is the most limiting factor to the
technique. In FIB-SEM, whether imaging or milling, the ion beam introduces Ga
þ
ions on/into the sample as well as destroying its surface [29]. These positive gallium
ions reduce the charging affect of biological samples, having a similar effect as gas
in variable pressure-SEM, to neutralise the charge at the sample surface [29, 30].
The SBEM, while it does not give as good of a lateral and axial resolution as
FIB-SEM, still has many advantages. The removal of material by a diamond knife
means the process is very fast, taking only a few seconds to remove a section from
the entire block-face. Nevertheless, this is a limiting factor, since the size of the
sample is restricted to the width of the diamond knife (typically 1.2–1.5 mm) and
should be as thin as possible to reduce the charging volume. The FIB only allows
milling of a relatively small area (in the range of 50 Â 50 µm) at any one time and
the area of this exposed face dictates the speed at which the cutting can take place.
Thus, this technique is well suited to smaller volumes and achieving high resolutions. Titze and Genoud [10] give a comparison of the best voxel sizes achieved
(x, y, z): 10 Â 10 Â 25 nm in SBEM, [31] and 5 Â 5 Â 5 nm for FIB-SEM
[32, 33]. However, Russell et al. [9] have shown that they could produce a large
dataset cutting slices at 10 nm and thus giving isotropic resolution for the SBEM of
10 Â 10 Â 10 nm.
The interplay between resolution and size is a problem common to all imaging
techniques [34]. The best image quality from the region of interest (ROI) would be
achieved by a small pixel size, high dwell time (the amount of time that the beam
will collect data at each pixel) and often a large FOV. However, this is rarely
achievable at the same time. The stability of the sample and the system in use will
5 Volume Scanning Electron Microscopy: Serial Block-Face …
121