Chapter 10 section 5 Enrichments: X-ray Dot Mapping
10.5er Analog X-ray Area Scanning (Dot Mapping)
10.5er.1 Procedure
X-ray area scanning is performed by the same technique as conventional SEM
imaging: the beam is scanned on the specimen in a raster pattern in the usual fashion
and the synchronously-scanned CRT is intensity modulated with the signal derived from
the detector, in this case either the energy dispersive or wavelength dispersive x-ray
spectrometer. The difficulty with this approach is the low intensity of the x-ray signal.
Compared to the backscattered electron signal, the x-ray signal from a particular
element is lower by a factor of 10,000 to 100,000 or more. The x-ray signal is generally
too weak, except for the case of a very high concentration (> 25%) constituent for which
there exists an efficiently excited x-ray line, to allow the depiction of a continuous gray
scale (such as we could derive from the x-ray ratemeter signal). During the scanning
process in the dot mapping mode, whenever an x-ray is detected in a pre-selected
energy range by the x-ray spectrometer, the corresponding beam location on the CRT is
marked by raising the intensity of the CRT beam to full brightness (white). A "dot" is thus
created on the display to note the detection and mark the location of the x-ray event, and
this display is recorded photographically. As shown in Figure 10.5er.1, the image thus
created has the appearance of a field of fine white dots, hence the term "dot mapping";
the procedure is also referred to as "area scanning".
To produce acceptable results, such as those illustrated in Figure 10.5er.1, the
dot mapping procedure requires careful attention to the instrumental operating
conditions:
(1) X-ray excitation:
We wish to maximize the count rate consistent with the spectrometer
characteristics, so we choose a sufficiently high beam energy to achieve an overvoltage
of at least U = 2 for the x-ray line(s) of interest, and a beam current current to produce
the maximum acceptable deadtime, e. g., 30 - 40% with an EDS or 100,000 cps for a
WDS on a bulk pure element standard. The elements of interest in the unknown will be
at lower concentrations and will produce proportionally x-ray lower count rates. For
WDS, we could choose to increase the beam current on the unknown still further, since
all other x-ray peaks from the specimen are excluded by the diffraction process. For
EDS, this option is not available, because the limiting deadtime is determined by the
entire x-ray spectrum, rather than just the peak of interest for mapping.
(2) Number of dots required for high quality maps
In conventional SEM imaging with electron signals, a photograph is normally
recorded with a single sweep of the scan with a frame time of 50 to 100 seconds.
Because of the low x-ray signal rate, the image accumulation must be made for
substantially longer times. Experience has shown that to obtain high quality dot maps,
105 -106 x-ray pulses must be accumulated over the entire image field, with the optimum
number highly dependent on the uniformity of the spatial distribution of the constituents
being mapped (Heinrich, 1975). A good procedure is to collect for total counts rather
than total time, scanning continuously over the frame and using a running integration on
the x-ray spectrometer to determine the cutoff point. An added advantage of this
approach is that any changes in operating conditions during the long time necessary to
accumulate the dot map will be integrated evenly over the entire recorded image, rather
than producing a noticeable drift in the dot map which would be the case for a single
frame exposure.
Achieving satisfactory results with dot mapping often requires trial and error.
Figure 10.5er.2 illustrates a typical dilemma. In Figure 10.5er.2 (a), accumulation of
10,000 dots gives good contrast in the higher concentration areas, but when 40,000 dots
are accumulated to develop contrast in lower concentration areas, the higher
concentration areas are completely saturated. (Note: The calcium in Figure 10.5er.2 is
localized over a small portion of the image. If this constituent were more evenly
dispersed over the image, proportionally more counts would have been needed to
produce the contrast shown.)
(3) Accumulation time for dot mapping
A considerable time penalty must be paid to achieve high quality dot maps. For
the optimum case of a wavelength dispersive x-ray spectrometer and a high probe
current of 100 nA or more, the count rate from a pure standard can reach 5x10 4 - 1 x 105
counts/second. If the element of interest is present in the unknown at a level of 10
weight percent, the count rate will be reduced to 5x10 3 counts/second, neglecting matrix
effects, and an integrated count of 106 x-rays can be accumulated in 200 s. For a
constituent present at the 1 weight percent level, the scanning time would have to be
increased to 2000 s. If the x-ray line is less efficiently excited or the beam current is
reduced, the accumulation times must be increased accordingly.
For energy dispersive spectrometry, it is important to carefully choose the
proper EDS pulse processing time (or amplifier time constant, see Chapter 7). If the
peaks of interest are well separated in the EDS, then we can accept a poorer resolution
to obtain a higher limiting count rate. The typical limiting count rate, even with the
poorest resolution (longest amplifier time constant), is about 10,000 cps, if we wish to
avoid possible count rate artifacts, described below. For EDS, this count rate applies to
the entire energy range of the incident beam, since photons from all excited peaks and
the x-ray continuum all contribute to the deadtime. The count rate from a constituent
present at the level of 10 weight percent may only be 100 - 500 cps, depending on the
excitation of other elements in the specimen, so the accumulation time for 10 5 x-rays will
be 2000 to 10,000 seconds. We therefore tend to accept lower quality maps with EDS
because of these count rate limitations. For concentrations lower than 10 wt%, the
accumulation times can become prohibitive. If the spectrum is complex so that higher
resolution and shorter pulse processing times are used, the limiting count rates will be
lower, and accumulation times longer.
(4) Adjustment of the CRT conditions:
To achieve the best dot maps, the analyst must carefully adjust the recording
conditions of the CRT, usually labelled "brightness" and "contrast". The ideal "dot" just
reaches the level of saturating the CRT phosphor. Such a dot is very fine and near the
resolution limit of the human eye, and when an adequate counts are accumulated, the
resulting dot maps show good spatial resolution. However, if the constituent is present
at a low concentration, it may not be practical to spend the time to accumulate sufficient
counts, and the dot map prepared with such fine dots will appear dark. Under such
circumstances, the CRT brightness and contrast are often increased so that the dot
"blooms" and enlarges, becoming more readily visible. Such an image appears brighter,
but the spatial resolution is degraded. These effects are illustrated in Figure 10.5er.2 (c)
and (d), where a consequence of brighter, larger dots is seen to be a loss of resolution
and contrast.
10.5er.2 Limitations and Artifacts
While dot mapping is a useful technique, it is subject to several significant
limitations:
(1) Qualitative, not quantitative information:
The dot map is only qualitative in nature, conveying the spatial location of
constituents but not the amount present. The critical information for quantitative
analysis, namely the count rate at each point in the image, is lost in the dot map
recording process. If any particular pixel is examined, the analytical information at that
pixel consists of either "present" (white dot) or "absent" (no dot). Since no gray scale
information exists, quantitative information cannot be represented. The area density of
dots, as seen in Figures 10.5er.1 and 10.5er.2, obviously suggests the variation in
composition, but the information is useful only when comparing areas within an image
which are much larger than single pixels. When a small area is examined, the noise in
the map overwhelms changes due to specimen composition.
(2) Sequential recording:
Dot maps are usually recorded for one constituent at a time directly on film
media. Although parallel recording is feasible, most systems only contain one photorecord CRT. Considering the time required to dot map a single constituent, sequentially
mapping for multiple constituents can impose a prohibitive time investment.
(3) Lack of flexibility for subsequent processing:
Recording on film media greatly reduces the flexibility for subsequent
processing of the information. For example, registration of multiple images for color
superposition on film is difficult.
(4) WDS defocussing/EDS decollimation:
To prepare a dot map, the beam is typically scanned in a raster pattern which
carries it off the coincident point of the optic axis of the microscope and the WDS
spectrometer focussing or the EDS collimation axes. For WDS, as shown in Figure
10.5er.3, a consequence of this beam motion is that the x-ray source at the specimen
moves off the Rowland circle, so the Bragg condition for diffraction is not maintained
during the scan. Because x-ray diffraction peaks in the WDS are quite narrow in angular
range, the detected intensity will fall sharply as the Bragg condition is lost, a condition
called "defocussing", shown in Figure 10.5er.4 (a) for a vertical spectrometer. However,
the Bragg condition is maintained for beam positions on the specimen which are parallel
to the width of the diffracting crystal. A band of constant transmission is thus seen in a
dot map, with the intensity falling rapidly on either side. This defocussing effect
becomes more pronounced as the magnification is reduced and the scan excursion
becomes greater, becoming extremely noticeable below approximately 500x. The
defocussing artifact can dominate images and in some cases actually overwhelm the
compositional contrast in the dot map, as shown in in Figure 10.5er.5. The aluminum
and iron dot maps show an uneven intensity distribution that interferes with the true
contrast, even for these major constituents. In the iron map, the spectrometer
transmission is a maximum along the diagonal indicated. The two iron-containing
phases each have nearly uniform composition, but because of defocussing the contrast
between the phases is only satisfactory to show fine details of the interface between the
phases in the lower middle portion of the image, and is nearly lost at the top and bottom.
A practical solution to defocussing is to move the band of maximum
transmission to coincide with the scan position. In practice, the scan is rotated to bring
the scan line parallel to the spectrometer focussing band, and the band is moved
synchronously with the frame scan by rocking the spectrometer crystal (Cameca
Instruments, 1974). The resulting map is uniform over much greater distances (lower
magnifications), as shown in Figure 10.5er.4 (b).
EDS spectrometers operate on the line-of-sight and are not focussing devices,
so the defocussing artifact does not exist. However, the EDS system is commonly
equipped with a collimator to restrict unwanted collection of remotely generated x-rays.
This collimator typically limits the view of the specimen to a few millimeters laterally from
the optic axis. If the beam is moved sufficiently far off-axis to achieve low
magnifications, < 50x, the solid angle of collection of the spectrometer will diminish
relative to that obtained when the beam is located on the optic axis, and the collection
efficiency will decrease, producing a drop in intensity in an x-ray dot map.
(5) Count rate effects:
The long pulse processing time of the energy dispersive spectrometer leads to
significant deadtime effects. As shown in Chapter 7, the EDS output is paralyzable due
to deadtime. That is, as the rate of photons arriving at the detector increases, the output
of discrete pulses from the detector/amplifier first increases linearly, but eventually a
maximum output count rate is reached beyond which the output actually decreases with
further increases in the input photon rate. If a sample consists of two phases with quite
different concentration levels, then the x-ray count rate from areas with the high
concentrations can be thought as occupying two linearly scaled locations along the input
pulse rate axis. However, because of deadtime, if the beam current is so high that the
input count rate from the high concentration phase is "over the hump" of the deadtime
response curve, then the high concentration phase will produce fewer output pulses than
the low concentration phase, reversing the apparent chemical contrast in the dot map.
In practice, even more complicated effects may be encountered. Because the EDS
detector deadtime is created by photons of all energies reaching the detector, the actual
output pulse rate from a preselected energy window is actually susceptible to count rate
effects from the whole spectrum, including both characteristic peaks and continuum
background.
(6) Lack of a background correction - detection limits:
In the conventional analog dot mapping procedure, no background correction is
possible since any x-ray in the acceptance window of the spectrometer must be counted.
The limit of detection in the dot mapping mode is poor. Characteristic and continuum xrays which occur in the energy acceptance window of the WDS or EDS spectrometer are
counted with equal weight. With WDS, detection limits in the dot mapping mode are
approximately 0.5 -1 weight percent, while for EDS systems, which have a much poorer
peak-to-background, the limit is approximately 5 weight percent.
(7) Lack of a background correction - continuum image artifact:
A second consequence of not performing a background correction is the
susceptibility of dot mapping to an image artifact in which compositional contrast may
seem to exist when an element is not actually present in the sample! This artifact can
occur in maps where a background correction is not applied because of the dependence
of the x-ray continuum on the average atomic number of the target (Kramers' expression,
Chapter 6). Thus, if a specimen consists of phases of different composition, the
background at any x-ray energy will vary proportionally to the local average atomic
number. This effect can be ignored when mapping major constituents (> 10 weight
percent), but the continuum artifact can dominate minor or trace level maps. Because of
the poorer peak-to-background ratio, this artifact is more significant for EDS
spectrometry than for WDS. A typical EDS P/B ratio for the FWHM measurement
window is in the range 30/1 to 50/1. At mass concentration levels of 1 - 3 wt%, the
background will be contributing as much intensity as the characteristic peak. If the
analyte is present at the same concentration in two phases whose average atomic
numbers differ by a factor of two, there will be an apparent doubling in the concentration
of the analyte due to the change in continuum between the two phases. An example of
such false contrast due to the continuum artifact is shown in the EDS compositional
shown in Figure 10.5er.6. The specimen is aluminum - copper eutectic. The maps have
been prepared with the energy dispersive x-ray spectrometer, with an energy window at
the position of aluminum, copper and scandium. The contrast in the scandium image
suggests that scandium is present and is preferentially segregated to the copper-rich
phase. In reality, there is no scandium is present and the false contrast is entirely due to
the atomic number dependence of the continuum. When a proper background
correction is applied, the apparent scandium contrast is eliminated. (Myklebust et al.,
1989). For WDS with its superior peak-to-background, the continuum artifact becomes
significant below 1 weight percent.
(8) Poor contrast sensitivity at high concentration levels:
The dot mapping technique has poor contrast sensitivity, particularly at high
concentration. The quality of a dot map depends on the concentration level of a
constituent. With sufficient time expenditure, it is possible to prepare a dot map of a
region containing a constituent at 5 weight percent against a background which does not
contain that constituent. However, it is virtually impossible to visualize a 5 weight
percent increase in concentration above a generally high concentration level, for
example, 50 weight percent. A high density of dots is produced by the high
concentration level of the analyzed element, reducing the visibility of the contrast
resulting from the small change in concentration, as illustrated in Figures 10.5er.2 (a)
and (b).
10.5er.3 Dot Mapping at Extreme Limits
Figure 10.5er.7 shows an example of dot mapping zinc at the grain boundaries of
polycrystalline copper. The levels of zinc being mapped, as confirmed by
separate point analysis and by digital compositional mapping range from 0.1 to
10 weight percent. This map required over 6 hours of scanning at a beam
current of 200 nA to reach these low concentration levels. In this case dot
mapping is successful because there is no zinc present except in the areas of
interest, so that the compositional contrast is very high, even at these low
concentrations.
Figure 10.5er.8 shows an even more extreme example of this case of dot
mapping low concentration levels. The dot map shows trace level imaging of
aluminum in human brain tissue where the maximum level of aluminum is 500
parts per million (Garruto et al., 1984). To reach these levels, the map required
an accumulation time of 15 hours at high beam current (500 nA). Again, dot
mapping was successful because the constituent of interest, aluminum, was only
present in the region of interest and was at much lower levels elsewhere in the
structure.
References
Cameca Instruments (1974). Users Manual for CAMEBAX.
Garruto, R. M., Fukatsu, R., Yanagihara, R., Carleton Gajdusek, D., Hook, G., and Fiori,
C. E., (1984). Proc. Nat. Acad. Sci. (USA), 81,1875.
Heinrich , K. F. J. (1975). "Scanning Electron Probe Microanalysis" in Advances in
Optical and Electron Microscopy, Barer, R. and Cosslett, V.E., eds., 6, 275.
Myklebust, R. L., Newbury, D. E., and Marinenko, R. B. (1989). Anal. Chem.
61,1612.
Figure Captions
10.5er.1 X-ray dot map (area scan) of a ternary alloy: (a) Specimen current
(inverted contrast) image showing atomic number contrast; (b) dot map of silver
(L); dot map for copper (K); (c) dot map of tin. The dot maps have been
prepared with the characteristic x-ray signal derived from a wavelengthdispersive x-ray spectrometer.
(Example courtesy Robert Myklebust, NIST.)
10.5er.2 Choice of recording parameters in x-ray dot mapping:
(1) Effects of number of dots: (a) Good contrast is found in high concentration
region with 10,000 dots recorded; (b) further accumulation to 40,000 counts
leads to improved contrast in lower concentration area at the expense of
saturation in higher concentration region.
(2) Choice of dot brightness: (c) properly adjusted CRT dot; (d) dot adjusted to be
too bright; note loss of contrast in image. Specimen: coating on steel; E o = 20
keV. (a),(b) = Ca K(c),(d) = Fe K(Example courtesy Ryna Marinenko, NIST).
10.5er.3 WDS defocussing during mapping. As the beam is scanned off the
optic axis of the WDS, the displacement in the specimen plane is equivalent to a
change in the angle away from the Bragg angle, B , of the x-rays approaching
the crystal by an angle .
10.5er.4 (a) Example of a dot map with defocussing evident. The band of
intensity corresponding to the exact focus. The band axis is parallel to the width
dimension of the diffraction crystal. Aluminum K from a geological sample. (b)
Correction of the defocussing by dynamic crystal rocking. The diffraction crystal
was rocked during the scan to move the band of constant intensity from top to
bottom during the scan. (Cameca Instruments, 1974).
10.5er.5 Example of the WDS defocussing artifact as seen in dot maps of an
iron-aluminum electrical connection interface. The arrow marks the axis of the
defocus band. (a) aluminum K map (b) iron K map. Note the non-uniformity
of the intensity and the loss in contrast near the bottom of the aluminum map;
note also the different orientation of the focus bands in the two maps because of
the different positions of the wavelength spectrometers relative to the specimen.
10.5er.6 Example of false compositional contrast, a mapping artifact which
occurs because of the dependence of the x-ray bremsstrahlung on the specimen
composition. The specimen is aluminum - copper eutectic. A map has been
prepared with the energy dispersive x-ray spectrometer, with an energy window
at the position of
(a) aluminum; (b) copper; and (c) scandium. The contrast in image (c) suggests
that scandium is present and is preferentially segregated to the copper-rich
phase. There is no scandium is present. (d) When a proper background
correction is applied, the apparent scandium contrast is eliminated. (Myklebust et
al., 1989).
10.5er.7 Dot map of zinc at the grain boundaries of polycrystalline copper. The
compositional map reveals concentrations as low as 0.1 weight percent (sample
courtesy Daniel Butrymowicz, NIST).
10.5er.8 Trace level imaging of trace constituents in human brain tissue:
aluminum, maximum level = 500 parts per million (Garruto et al., 1984).