вход по аккаунту



код для вставкиСкачать
Cross-Section TEM Sample Preparation of Multilayer
and Poorly Adhering Films
Department of Materials and Metallurgical Engineering, Queen’s University, Kingston, Ontario, Canada
aperture grid; mechanical polishing; silicon; transmission electron microscopy;
TEM specimen preparation
The preparation of TEM cross-section samples from multilayer films or poorly
adhering films is discussed in detail in a step-by-step approach designed to enable a competent
experimentalist to reproduce the technique. The samples are mounted on an aperture grid and
mechanically polished to 2–3 µm in thickness. After ion beam milling for a short period of time (less
than 1 hour), a large electron transparent area is obtained. Examples from several thin film systems
are discussed. Microsc. Res. Tech. 36:368–371, 1997. r 1997 Wiley-Liss, Inc.
Mechanical thinning, dimpling, and ion milling are
the standard techniques employed to prepare crosssection TEM samples of thin films. However, in a
multilayer film particularly where one of the constituents is significantly denser than the others, for example, lead zirconium titanate on platinum on titanium on silicon oxide on silicon, it is difficult to obtain a
thin sample from the full depth of the film which may be
up to 10 µm thick. In the same way, films which
delaminate from the substrate cause considerable
sample preparation problems using the conventional
raft technique (Chew and Cullis, 1987). Generally when
the sample is thinned during the dimpling stage, the
disc separates along the central glue line as the film
delaminates. In order to overcome this problem, a
two-bar raft is made, which is epoxied to an aperture
grid, then mechanically thinned. In this way, minimal
stress is placed on the film and the grid provides
mechanical support. The sample is then thinned in a
parallel manner, similar to that in the tripod method
(Anderson et al., 1990). Since a uniform thin sample is
produced, the problems associated with multilayer
films are also overcome. This technique was specifically
developed for silicon-based materials in the microelectronics industry but it is easily extended to other,
non-semiconductor substrates.
The basic technique is shown schematically in Figure
1. A roughly square piece of material approximately 2 3
2 cm is removed from the bulk sample. If the original
substrate is silicon, these pieces are epoxied together,
film sides facing, with an opaque epoxy such as Epotek
H-22 (Epoxy Technologies, Billerica, MA); if not, a piece
of silicon is substituted for one of the original sample
pieces. The silicon acts as a thickness monitor as
described below. The pieces are then cut into bars 2 3 3
mm. The bar is mounted onto a lapping stub using a
heat sensitive wax (available from Gatan Inc., Pleasanton, CA) or low melting point polmer (such as Crystalbond 509, Aremco Products Inc., Ossining, NY) and
mechanically thinned to about half the original thickness. Mechanical thinning is carried out by placing the
stub in a Disc Grinder (Gatan Inc., or E.A. Fischione
Instruments Inc., Export, PA), which produces parallelsided discs initially using 240, then 600, grit paper. The
weight of the Disc Grinder also ensures that when the
sample is transferred to a polishing wheel and thinned
using a succession of diamond impregnated papers in
the order 30, 15, 9, 6, 3, 1, 0.5, 0.1 µm and a final polish
using Syton (Remet Chemical Co., Chadwicks, NY) on a
flocked twill cloth, the sample remains parallel sided.
At each paper change, it is necessary to examine the
sample under a zoom light microscope to ensure all
polishing marks from the previous paper have been
removed before proceeding to a finer grade. After the
final polishing step, the sample should have a mirror
finish with no polishing marks visible. The sample is
then removed from the stub and cleaned in acetone to
remove the wax, then mounted polished side down onto
an aperture grid, as shown, with M-Bond 610 (M-Line
Accessories Measurements Group Inc., Raleigh, NC) or
a similar transparent epoxy. The size of the aperture
can vary, however, an 800-µm aperture has proved to be
the most successful. It is very important that the
aperture should be free from epoxy and the sample is
centred. After curing, the sample is placed on a glass
lapping stub with heat sensitive wax and the same
mechanical thinning procedure is followed. The 240
and 600 grit papers are again used to produce a flat,
smooth sample. After transferring to the polishing
wheels, the sample must be frequently monitored with
a zoom microscope both with incident light to check for
scratches and with back lighting to monitor the sample
thickness. It is possible to monitor the sample thickness
with a metallurgical microscope fitted with a vernier,
but using back lighting is simpler and quicker. The
sample should be thinned on the 30-µm paper until
light is visible through the centre epoxy line, which at
*Correspondence to: Dr. Louise Weaver, Department of Materials and Metallurgical Engineering, Queen’s University, Kingston, Ontario, Canada K7L 3N6.
Received 1 December 1994; Accepted in revised form 16 March 1995.
Fig. 1.
A schematic representation of the sample preparation technique.
this stage has thinned sufficiently to become transparent. At this point the sample is transferred to the 15-µm
paper and thinned until the aperture is just visible as a
red circle through the silicon. In this way, the silicon
can be used to monitor the thickness of the sample
throughout the whole polishing procedure. The sample
is then transferred to the 9-µm paper and thinned until
the silicon is orange. The sample is then polished down
through each paper to remove the polishing marks from
the previous paper and final polishing is carried out
with Syton. It is very important to continuously monitor the sample thickness using the silicon as a measure.
After polishing, the sample will appear yellow with
back lighting and will be 2–3 µm thick. The stub is
placed on a hot plate to allow the wax to melt, then the
grid is removed by sliding it off the stub; if it is lifted off
the stub the surface tension of the wax will rupture the
thin area in the aperture. It is also possible to ‘‘float’’ the
specimen off in an acetone bath, which presents less
possibility of damage. The sample is placed in acetone
to remove the residual wax, then transferred to a fresh
acetone bath and left for 1–2 hours. The sample can
now be ion milled, preferably using a low angle, but will
require only a very short time (less than 1 hour) on the
ion mill.
Using this technique, the electron transparent area
is the whole of the 800-µm area of the aperture. Figure
2 shows a multilayer film of lead zirconium titanate
Fig. 2.
Fig. 3.
A lead zirconium titanate film on Pt/Ti.
A poorly adhering copper film on a silicon substrate.
Fig. 4. An as-deposited ruthenium oxide film on a silicon substrate coated with a blanket layer of
silicon oxide.
(PZT) on platinum on titanium (which has reacted with
the platinum) on silicon oxide in which each layer is
clearly visible. The density and thinning difference
between each layer is not reflected in this micrograph.
An example of a poorly adhering film is shown in Figure
3. The as-deposited copper layer is barely in contact
with the substrate and was impossible to make using
conventional sample preparation techniques. The ruthenium oxide film shown in Figure 4 is over 5 µm thick
and is obviously electron transparent over the whole
micrograph. All the micrographs were obtained using a
Philips (Mahwah, NJ) CM20 microscope.
sample, which can then be ion milled in the conventional way.
A technique to produce cross-section TEM samples
from multilayer or poorly adhering films has been
described. This technique utilises standard metallurgical polishing techniques to produce a 2–3-µm-thick
Anderson, R.M., Klepeis, S., Benedict, J., Vandygrift, W.G., and
Orndorff, M. (1989) Recent developments in the preparation of
semiconductor device materials for the transmission electron microscope. Inst. Phys. Conf. Ser., 100:491–500.
Chew, N.G., and Cullis, A.G. (1987) The preparation of transmission
electron microscope specimens from compound semiconductors by
ion milling. Ultramicroscopy, 23:175–198.
The laboratory assistance of David Mayer of Northern Telecom, Ottawa, was invaluable in developing this
technique. Thanks are also given to Dr. Lynnette
Madsen of Linkoping University, Sweden, and Ellen
Griswold of Queen’s University, Canada, who provided
the samples that required this technique to be developed.
Без категории
Размер файла
232 Кб
Пожаловаться на содержимое документа