Aug 10, 2004 - Note that the following instructions are not suitable for the cross-section .... Using a GATAN 656 dimple-grinder, a sphere-shaped deepening is ...
Preparation of cross-section samples for transmission electron microscopy (TEM) Dr. Elisabeth Mu ¨ ller & Daniel Abou-Ras
Elektronenmikroskopie-Zentrum der ETH Z¨ urich (EMEZ) c/o Institut f¨ ur Angewandte Physik ETH-H¨onggerberg CH-8093Z¨ urich
Contents 1 Cutting slices
1
2 Forming a stack
3
3 Embedding the stack in a tube
5
4 Cutting the tube into disks
7
5 Polishing the disks
9
6 Dimple-grinding
11
7 Ion milling
14
8 Final TEM sample
17
i
Preface ! Note that the following instructions are not suitable for the cross-section preparation of every kind of sample material; if your sample is soluble in water or acetone, you should rather consider a sample preparation by e.g. focused ion beam (FIB) cutting! ! Also, this manual is written for cross-section sample preparation of samples that consist of a substrate with a stack of thin-film layers on top of it. It can be easily adapted to the preparation of e.g. metal samples as well.
Z¨ urich, August 10th, 2004
ii
Chapter 1 Cutting slices Step-by-step: - slices of 2 mm width and 4 - 6 mm length are cut out of the appropriate sample using a diamond wire saw; if the sample material consists of cleavage planes as e.g. Si, it is possible to cut the slices using a diamond scriber, e.g. from DIAMOND SA (http://www.diamond-fo.com, phone +41 91-785 45 45) - slices of about 2 mm width and 4 - 6 mm length are cut out of a Si(001) waver along the [110] directions. The best way to do this is using a diamond scriber (see above) - since the slices will finally be embedded in tubes with an inner diameter of ca. 2.12 mm, the width of the slices must be smaller than this. The maximum length of the slices is given by the length of the tubes
1
1. Cutting slices
Figure 1.1: Cutting slices of 2 mm width out of a thin-film solar cell
2
Chapter 2 Forming a stack There are several ways to form stacks, which will then be embedded in a metal tube. The procedure depends strongly on the thickness of the substrate of the sample: - you may glue two slices of the sample together so that the layers face each other - you may glue a slice of a Si wafer on top of a slice of the appropriate sample (with Si on top of the layers) Note that in both cases, the total stack thickness must be smaller than the inner diameter of the metal tube (ca. 2.12 mm). To achieve the appropriate thickness, the substrate of the sample has probably to be ground, which could introduce cracks within the thin films. Thus, if the substrate thickness of the sample is about 1 mm, and also if only little sample material is available, the second method to form a stack should be considered. In general, the layers of interest should be situated in the middle of the stack, so that they are in the center of the final TEM sample. For gluing, GATAN two-component epoxy glue is used. The standard ratio between hardener and resin is 1:10. If the sample material is brittle, a ratio of down to 1:12 may be preferable. At 120 ◦ C, the hardening time is about 2 minutes.
Step-by-step: - for mixing the glue, cut a piece of Al foil with scissors, fold piece four times - put appropriate number of droplets of both hardener and resin on the Al foil - stir with toothpick 3
2. Forming a stack
- put slice of Si wafer on another piece of Al foil - apply epoxy glue sparingly on the slice of Si wafer - put slice of sample on top of slice of Si wafer - anneal at 120◦ C; the best way to find out when the epoxy glue has hardened is to put a small drop of epoxy glue next to the stack on the Al foil - in order to have as little epoxy glue as possible between the slices (for better stability of the sample, less preferential etching of the epoxy glue during ion etching and applicability of the auto-stop in the ion etcher) the slices are pressed during annealing of the epoxy glue. If the pressure, however, is only a little bit too strong, this will again lead to cracks in the sample material. Therefore, it is recommended to put some pressure onto the stack with the tweezers before annealing and only fix the stack during the annealing
Figure 2.1: Left: Mixing hardener and resin in a ratio of 1:10. Right: Forming a stack by slice of solar cell on the top and Si wafer on the bottom
4
Chapter 3 Embedding the stack in a tube After annealing the stack for a few minutes, it is ground so that it can easily be introduced into a tube with an outer diameter of 3 mm and an inner diameter of ca. 2.12 mm. This is done to stabilise the TEM sample. The stack is embedded in the tube again with the GATAN two-component epoxy glue. The tube can be composed of various materials. They should neither be magnetic nor contain one of the elements to be detected e.g. by EELS or EDX. Typically the tubes are made of copper, steel or some ceramic material.
Step-by-step: - check whether the stack fits into the respective tube; if it does not, grind the stack with abrasive paper - clean the tube in an acetone bath - place the tube standing on a piece of Al foil - fill a droplet of the epoxy glue into the tube (use toothpick) - drop the stack into the tube on top of the epoxy glue - place the tube standing on the Al-foil on a heating plate (120◦ C); the epoxy glue becomes more fluid after a few seconds, the stack will sink into the epoxy glue - only if the stack does not sink totally into the epoxy glue, a small pressure by the tweezers may be applied - after the initial droplet of epoxy glue has hardened, the tube is filled with some more epoxy glue: always add droplets where the distance between the 5
3. Embedding the stack in a tube
stack and the wall of the tube is smallest. This way, you will avoid gas bubbles within the adhesive, which would destabilize the samples during further preparation - some epoxy glue should always run out of the tube - this tells you the tube is completely filled with glue; the colour of the epoxy glue will change from transparent-yellow to brown when it is fully hardened; you may also check with the tweezers
Figure 3.1: Left:Filling a droplet of the epoxy glue into an Fe tube. Middle: Dropping the stack into the tube. Right: The stack is sealed within the hardened epoxy glue.
6
Chapter 4 Cutting the tube into disks For cutting disks out of the tube, Al holders with an appropriate notch for the tubes are used. The tube is glued onto the Al holder with thermo-wax and cut using a diamond wire-saw.
Step-by-step: - the tube is glued with thermo-wax on an Al holder. With a diamond wire saw disks of about 300-400 µm thickness are cut - when mounting the tube on the Al holder, put thermo-wax all around the tube, thus, best adhesion to the Al holder is guaranteed, and also the cut disks will not drop off the holder - keep in mind that the wire of the saw has typically a thickness of about 300 µm; therefore, a distance of 600 - 700 µm from one cut to the next is reasonable - the counterweight applied on the wire depends on the stability of the sample material. The more brittle it is, the less counterweight should be applied - the speed of the diamond wire saw should be kept rather low - one Al holder may be used to cut about 20-30 disks, i.e. several tubes! - just cut once at each position of the Al holder! - if the sample contains a layered structure which easily delaminates, the tube has to be placed with the interfaces parallel to the wire. Best fix the tube in such a way onto the Al holder that the Si-wafer on top of the layered structure faces the diamond wire (see figure 4.1) 7
4. Cutting the tube into disks
Figure 4.1: Fe tube mounted on the Al holder: the Si wafer faces the diamond wire
8
Chapter 5 Polishing the disks Polishing the disks down to a thickness of about 80 - 120 µm may be performed by hand or using a lapping machine, on wet grinding paper, depending on the brittleness of the sample material. In figure 5.2 optical micrographs of an unpolished and a polished disk are shown.
Step-by-step: - select a glass cylinder to be used as a support of the disk and measure its height using a micrometer gauge - glue one disk with thermo-wax on top of the glass cylinder (see figure 5.1) - polish one side plan-parallelly to the other side - flip the disk on the other side - measure the total thickness (cylinder+disk); calculate the disk thickness - polish the other side; the disk is to be ground down to a thickness of about 95 - 110 µm; thus, one achieves high stability as well as short time for dimplegrinding - at the lapping machine, use first 400-, 1000- or 1200-grade grinding paper (depending on the brittleness of the sample material, then 4000-grade grinding paper - if the polishing procedure is done by hand, move the glass cylinder with the mounted disk in circles on the grinding paper; the thickness of the sample may be estimated by the number of turns on the milling paper, if the diameter of the circles are always approximately the same. When polishing the second 9
5. Polishing the disks
side of the disk, twist the glass cylinder several times to prevent the disk from becoming wedge-shaped - the thickness of the disk may also be controlled by checking the colour of the epoxy glue - if the sample contains a layered structure which easily delaminates, always grind in a direction parallel to the layers! - remove the disk from the glass cylinder on the heating plate and check the final thickness of the polished disk using a micrometer gauge
Figure 5.1: Glass cylinder with disk mounted on the top
Figure 5.2: Left: unpolished disk. Right: polished disk
10
Chapter 6 Dimple-grinding Using a GATAN 656 dimple-grinder, a sphere-shaped deepening is ground into the polished disk. The dimple-grinder consists of a specimen mount, on which the glass cylinder with the polished disk is mounted, and a dimpling wheel. For the polishing procedure, a suspension of diamond particles in oil or in water is used (oil is preferable, since it does not evaporate at room temperature). Depending on the brittleness of the sample material, the disk may be dimple-ground just from one side or from both sides. Dimple grinding the sample from both sides reduces the deformation of the surface; for very brittle sample materials, dimple-grind just from one side. The depth of the deepening should not exceed ca. 80 µm for one-side dimplegrinding, and 40-45 µm for two-side dimple-grinding. The remaining sample thickness in the middle of the deepening should have 15-20 µm for silicon. For III/V semiconductors, 25-30 µm is preferable, as these materials are rather brittle. The deeper the deepening is, the shorter will be the ion milling process afterwards. As for the counterweight, in the case of silicon 30-40 g, for III/V semiconductors 20 g may be applied; in metals 10 g could already be too much, because in metals dislocations are introduced very easily. See figure 6.1 for details.
Step-by-step: - first, center the glass cylinder with the polished disk using the binocular in such a way that the desired position of the later hole remains in the middle of the turning specimen mount - adjust the counterweight of the dimple-grinder; for brittle samples begin with a small counterweight (e.g. 15-20 g), then increase it slowly.
11
6. Dimple-grinding
- stop turning of dimpling wheel and specimen mount - lower the cam control carefully (very left wheel on the side) - turn micrometer drive (left knob next to the specimen mount) counterclockwise, until dial indicator remains at zero (the zero point need not be identical to the zero on the dial indicator) - press twice ”zero” - turn table wheel clockwise to set desired thickness of deepening; the digital dimple depth display is more accurate than the analog dial indicator - raise the cam control - put two 2 drops of grinding suspension (enough!) on the sample - turn on the dimpling wheel (”arm”) and the specimen mount (”table”); best clean the dimpling wheel with acetone - carefully lower the cam control, lowering the turning dimpling wheel on the sample - from time to time, raise the cam control, clean carefully both dimpling wheel and disk, add drops of grinding suspension, and lower the cam control again - if too much suspension is added, it may run in the inner part of the magnetic turntable or the dimpling wheel, thus, hindering the dimple-grinding; in this case, stop grinding, and clean everything carefully with a wiping cloth - if the sample contains a layered structure which easily delaminates, the disk may be positioned in such a way that the interfaces are parallel to the dimpling wheel, without any rotation of the specimen mount in order to prevent delamination
12
6. Dimple-grinding
Figure 6.1: Setup of dimple-grinder
13
Chapter 7 Ion milling For the ion milling process, a GATAN 691 Precision Ion Polishing System (PIPS) is used. In this machine, two focussed Ar ion beams mill the dimple-ground sample in such a way that a hole results at the desired position. The deepening of the sample at its lower side with respect to the PIPS holder should not be deeper than about 40 µm since otherwise the ion beams etch too far away from the center of the sample. In general, the parameters for the ion milling process are rather specific for each material and have to be optimized. The ion milling rate increases with higher etching angle and higher etching voltage; however, the sample is also more severely damaged. Therefore, the angle as well as the voltage should be kept rather low. Generally, a higher voltage combined with lower angle is less harmful to the sample than lower voltage combined with higher angle. For a sample with a thickness at its edge of 100 µm, the lowest etching angle is about 2◦ , if the sample is dimpled from both sides, and about 4◦ , if it is only dimpled from one side. For silicon, 4◦ and 4.3 kV are good values. For (Ga,Al)As, 3◦ and 3.5 kV are o.k., while for CaF2 on GaAs, 2.5◦ and 3 kV should be chosen. For metallic samples values of 4-6◦ and 5-7 kV are often chosen. During ion milling the sample is severely heated; in the PIPS, temperatures clearly above 120◦ C can be reached. E.g., in samples containing GaAs, Ga droplets may appear; in this case, choose beam modulation instead of continuous etching. The beam modulation reduces not only the local temperature on the sample, but also prevents preferential etching of the adhesive between the crystal platelets and also of layers with higher etching rates, because these weaker materials are shaded.
14
7. Ion milling
Step-by-step: - mount the dimple-ground sample on the PIPS Al sample holder (see figure 7.1) in such a way that the desired position of the future hole is centered (use the binocular) - for sample dimple-ground on both sides, the deepening at the lower side has to be smaller than ca. 40 µm - for samples dimple-ground on one side, make sure that the dimpled side is oriented face up - make sure that the PIPS machine is turned on approximately 30 minutes before ion milling to reach stable, high-vacuum conditions - mount the Al sample holder into the PIPS - for layered structures, beam modulation should be used; make sure the layers are parallel to the marks on top of the PIPS - turn beam modulation on - close the cover, evacuate chamber - when the green light lights up, lower the sample holder into the ion milling chamber - set turning velocity, ion milling angle, ion milling voltage and duration of ion milling - start the ion milling process - the final hole should be as small as possible
Figure 7.1: PIPS sample holder
15
7. Ion milling
Figure 7.2: PIPS setup
16
Chapter 8 Final TEM sample A TEM sample is least contaminated directly after the ion milling process. Depending on the material, the samples often degrade relatively fast. In general, the samples are contaminated by hydrocarbonates, which can be removed by a short plasma-clean prior to the TEM investigation. Many of samples also degrade owing to oxidation or other chemical processes. In the latter case, only re-etching the sample in the PIPS can remove the contamination at the surface.
17
