Electron probe microanalysis Low Voltage SEM Operation Created

Electron probe microanalysis Low Voltage SEM Operation Created 9/22/10

What’s the point? Traditionally SEMs and microprobes operate at gun voltages (E 0) in the range from 15 -20 k. V. However, it is possible to operate at a wider range of accelerating voltages: down to around 1 k. V as well as up to 30 k. V. There are benefits under certain conditions of operating at these different (esp. lower) voltages.

“Thinking like an electron” You have seen with your Monte Carlo simulations that for a constant material, dropping the incident electron k. V value will decrease the scattering of the electrons in the sample. In fact this decrease goes approximately as a 1. 7 power, i. e. dropping from 15 to 1. 5 k. V will reduce the electron range (scatter) by 101. 7 which is a factor of 50!

Figure showing the SE 1 and SE 2 As seen in the figure SE 1’s occur immediately at the beam impact point. Of course, electrons continue to scatter within the sample and SE 2’s will emerge over a range of distances from that central point. Dropping the E 0 therefore should produce better resolution electron images.

There are potentially limits, however, to low voltage imaging: • Will the gun put out a “bright” enough beam at the particular lower voltage? • Is the sample surface clean? Going to lower k. V means that any junk on the surface will be preferentially enhanced in the image. -- On the other hand, if what you WANT to image is the junk on the surface, lower k. V is definitely called for! • SE detectors operate pretty well down to very low voltages (at or below 1 k. V). • However, many BSE detectors start to become less sensitive as you drop below 10 k. V. However, our Hitachi S 3400 BSE detector works very well below 5 k. V and even gives a weak image at 1 k. V.

EDS at Low Voltages In many cases, you are not just collecting images, but using EDS to qualitatively determine the composition of some phase in your sample. Operating at 15 or 20 k. V gives access to K lines of elements from B to ~Se, to L lines of elements from Fe to ~Au or Pb, and M lines of most of the rest of the periodic table. However, operating at say 5 k. V reduces the lines that are available for EDS examination -- and they are all crunched together, with potentially many interferences and non-unique interpretations.

Interferences at Low Voltages As Newberry (2002) points out, EDS operation at low k. V is fraught with difficulties, as demonstrated in his figure. If oxygen and/or carbon are present (either intentionally or not!), there are many important L and M lines that are overlapped. From Newberry 2002 Figure 6

Additionally, as the surface layers become more important (that’s the region the electrons are paying more attention to), then little details like oxide skins (most metals will form some oxide layer, even gold, according to one report I’ve seen). Therefore, a single low voltage EDS spectrum can be a convolution of both the deeper material composition plus the surface skin contribution -- which makes for nonunique solutions to the question: is there trace amounts of oxygen present in this metal?

Why do some say use high k. V for better images? A lot of books and folks with years of experience say that higher k. V gives better images. I assume there must be something to this, though I cannot say I’ve seen any side by side study of images from 5 k. V to say 30 k. V. For example, Goldstein et al, p. 197 go thru an explanation why a 30 k. V image of Silicon would be sharp at 100, 000 X -- for a 1 nm! resolution beam on a 1024 x 1024 pixel image (=1 nm pixels) the 1 nm SE 1 “signal” would have a high signal/noise ratio, with the SE 2 noise “being constant” and therefore presumably vanishing. However, nowhere do they explain why this is not the case also at say 15 k. V, and why 30 is better than 15 k. V.

Why do some say use high k. V for better images? Is it something inherent in 30 k. V? Is it that the SEM’s gun is “brighter” (more electrons in a tighter beam) at higher vs lower k. V? Is the detector more sensitive at higher k. V? My suggestion: be empirical… try going to higher, then to lower k. V, and see what you think is better…. and let me know what you decide!