Abstract
Silicon-drift detectors for energy-dispersive spectroscopy have gained increasing acceptance in the
field principally because of their outstanding capabilities for high count-rate performance coupled
with excellent resolution [1,2]. Improvements in design in the last few years have overcome some
early disadvantages of the SDD in comparison to Si(Li) detectors, which have dominated the field
for nearly 4 decades. Amongst these are detection efficiency in the low-energy regime, and a dropoff
in detection efficiency at higher energies above 10keV. The former disadvantage has been
largely overcome by generational design improvements to minimize the effective “dead” layer at the
entrance window, so that present detectors provide better performance in the low-energy regime than
Si(Li) detectors. In the latter case, higher energy peak detection performance has been improved by
increasing the thickness of the detector. This is important primarily for the use of the SDD in a
TEM environment, where very high count rates are not encountered due to the thin sample
geometry. Also of importance for incorporation of an SDD into a TEM column is the lack of need
for liquid nitrogen, since the SDD is chilled by a thermoelectric cooler. This is an advantage for
SEM and microprobe operation also, but high-resolution TEMs (particularly aberration-corrected
instruments) are more sensitive to vibration than SEMs or microprobes, so the much reduced form
factor of an air-cooled SDD relative to a Si(Li) detector with a large LN2 dewar is desirable for use
on the TEM. We have recently tested two Bruker C
