600 Wall and Simon
7. Most physical additives (which may improve the biological activity) such as
bovine serum albumin, trypsin, and high concentrations of peptides, oligonucle-
otides, or PEG cannot be used. One way to remove these is by passing the sample
of interest over an appropriate sizing column, if one can be found where it comes
off in the void volume. This can also be a useful way of exchanging salts or
buffers. If one is in a lab where bacteria are a problem, low levels (0.2%) of
sodium azide can be tolerated in buffer solutions. It seems to wash off grids and
not interfere with STEM samples.
8. Ca
2+
often does not wash off well, but Mg
2+
usually does. A buffer containing
Ca
2+
can be washed with Mg
2+
, then ammonium acetate.
9. Some salts and buffers are likely to cause problems. Phosphate buffers often do
not wash off well and leave bright spots, which interfere with the analyses. Some-
times, Tris buffers leave a bad background and they also interfere with glutaral-
dehyde fixation. A buffer control grid is made for a new sample (especially in an
unusual buffer) to see how well the components of the buffer wash off.
10. Carbon films are often made by shadowing onto cleaved mica. We have found
that films made on freshly cleaved NaCl crystals will float off onto a dish of
water much easier than off of mica. Mass analysis of the two background films
are indistinguishable, indicating that they are just as flat.
11. Many experimental details about the grids, films, and shadowing can be found in
Cells: A Laboratory Manual, Vol. 3, pp. 125.2–125.7 (CSH Press, 1998). How-
ever, to simulate STEM conditions for a sample, the only critical step is the thin
carbon film because it is the only surface that the biological materials will see.
Some additional capabilities of the STEM can be found in the same volume,
Chapter 124.
12. Caution: Steps taken to concentrate a sample, such as Centricon filters, may also
concentrate contaminants of the same size.
13. Many, but not all, molecules adsorb well to poly-lysine-pretreated grids. It
may be necessary to change the suggested concentrations to lower ones when
using them.
14. Biological specimens are sensitive to radiation damage by the electron beam. This
manifests itself in two ways, as a loss of mass and a loss of fine detail. For STEM
mass measurements at 40 keV with the specimen at –150°C, the rate of mass loss is
roughly 0.25% for every 1 el/Å
2
. The normal STEM imaging dose is 10 el/Å
2
,
which results in 2.5% mass loss per scan. Measurements are usually done on first
scan images, but it is instructive to do a dose-response curve using sequential scans
of the same area for a new specimen type. Most protein specimens plateau at approx
50% mass loss even at high dose. DNA has much less mass loss, whereas carbohy-
drate has much more. Resolution loss is seen most easily by examining the ends of
TMV rods. These have relatively sharp corners with one end concave and the other
convex. After several scans, the corners become rounded, providing a visual moni-
tor of the accumulated dose to that part of the specimen.
15. The expected error (standard deviation) for mass measurements arises from elec-
tron counting statistics and thickness variations in the thin carbon substrate. These