In order to determine whether the stage meets the set requirements , various experiments and tests regarding the applied principles and safe usage are proposed. While the main goal is to reach 1500°C, first, the other requirements should be tested to ensure the workings of the stage before pushing it to its limits. As the design has been split up into three requirement categories, the testing also follows these categories.
First, the influence of the geometry of the stage on the EBSD system should be tested. In the requirements it is formulated that a tilt range of 69 to 75 degrees must be achievable with a working distance of 25mm. These requirements are tested by inserting the stage and performing EBSD scans at one degree increments from 69 degrees to 75 degrees. This test is done at room temperature as there is no need for elevated temperatures to validate the working of the conical holes through
Page 20 which the electrons pass. These scans should be performed with flowing water in the cooling circuit to verify the influence of cooling on the stability of the system, especially with respect to vibrations.
To verify the thermal stability of the system, the remaining thermal drift of the thermal centre of the sample is investigated. Because the thermal centre of the sample is not a priori known, any displacement of a point can possibly be a combination of thermal drift and thermal expansion. To verify whether drift occurs, for three images taken at room temperature and two elevated temperatures, the thermal centre is constructed and tracked, the combination of drift and expansion can be monitored. The principle behind this monitoring is elaborated upon below.
Figure 11: Real and virtual thermal centres for an expanding sample. Where TC1, TC2, and TC3 represent the real thermal centres, the intersecting lines represent the virtual thermal centre, i.e. the point that appears to be stationary in the images. In the case of TC1 and no drift, as the real and virtual thermal centre coincide, it is concluded that no thermal drift
occurs.
In Figure 11 above, a schematic representation is given of how the movement of a real thermal centre and a virtual, constructed thermal centre behaves when a sample is heated. In the image, the green dots represent the unheated situation and TC1 is the location of the thermal centre of this sample. If no drift occurs, TC1 remains stationary when heated and the thermal expansion causes the indentations to move away from TC1 following uexpansion,ΔT1. The new location of the indentations is represented by the blue dots, lines drawn through the pairs of blue and green dots coincide at the location of the thermal centre TC1 as they should following the definition of a thermal centre [34].
Subsequently, if the temperature is increased, the indentations will travel further away from TC1
following uexpansion,ΔT2 as depicted in purple, but the lines drawn through their new locations and their initial location will still intersect at TC1. If this situation would be observed in experiments, it would prove that the sample is thermally centred.
Instead, if drift would have occurred for the same temperature increase, the thermal centre would no longer be positioned at TC1, but for example at TC2 depicted in orange following both uexpansion,ΔT1
and udrift,ΔT1. While the same thermal expansion occurred, the three indents have not only moved
Page 21 away the thermal centre, but have also drifted together with the thermal centre. The lines drawn through the pairs of orange and green dots still coincide but not in the location of the thermal centre TC2. While the thermal centre drifted to the left, the virtual thermal centre (constructed by the intersection of the lines through the new and initial location of the indents) has moved to the right.
Assuming this virtual centre to be correct and increasing the temperature results in the red situation following uexpansion,ΔT2 and udrift,ΔT2, where thermal expansion has increased and the real thermal centre travelled left to TC3 and the virtual thermal centre to the right. From the fact that the orange and red crossing of lines do not coincide, it can now be concluded that drift occurred.
Finally, if the various design requirements have been met, the temperature requirement of the stage should be tested. While monitoring the temperature of various components in the SEM, the sample and SiC-disk temperature is logged as a function of the power of the laser. The predicted temperature power relation can be found in Figure 12. As one of the possible applications of the heating stage is the research on tungsten at high temperatures and tungsten has a relatively low thermal conductivity [9], it is chosen to use a tungsten sample. With tungsten, it is shown that even for poorly conducting materials, the heating stage is able to reach the specified temperatures. While the power-temperature relation will be different for other materials, the test on tungsten validates the requirement of material independent heating.
Figure 12: Power temperature relation for a tungsten sample acquired by thermal simulations
In addition, to both validate the thermal simulations and ensure correct and safe working of the heating stage, the temperatures of various components of the heating stage should be monitored during heating by using thermocouples. As not all parts are sensitive to temperature, only those that may suffer from an increase in temperature are monitored. These are the heating stage frame, the front heatshield, the backside heatshield, and the tubing for the cooling circuit. If all parts remain at safe temperatures in steady state, it may be concluded that the heating stage is fit to conduct experiments of durations longer than 1 day.
3.1 Experimental results
The heating stage has been tested with a tungsten sample inside the vacuum chamber of the SEM at a power range of 0 to 30W. As the only available flange with a water feedthrough has the same diameter as the EBSD detector flange and only one feedthrough of this diameter is present on the vacuum chamber of the microscope, cooling and EBSD cannot be performed together at this point in time. No EBSD scans have been made using the heating stage since the water cooling must be active in order to safely reach elevated temperatures. However, the range of sample tilt angles is assumed
0
Page 22 to work correctly as imaging by secondary electrons (SE) was possible over the entire tilt range where the entire sample surface remained visible. This is shown in Figure 13 below where the black oval is the sample location. At each tilt angle, the entire sample surface remains visible to the electron beam. As the imaging parameters are the same for all three images, the decrease in contrast is caused by the increased difficulty for the electrons to reach the SE detector. This is due to the position of the SE detector.
Figure 13: SE images of the sample at minimal magnification where the sample remained visible for the entire range of tilt angles. Shown above from left to right are the extreme values of 69° and 75°, and the middle value of 72°.
Using the tungsten sample, the heating capabilities of the stage has also been investigated using the SE detector. During the initial testing of the dedicated heating stage, it has been found that the temperature had risen to a range in the vicinity of 1500°C. This is justified by looking at post-mortem EBSD scans of the sample and comparing them to those of a sample that has been machined out of the same plate as the heated sample and has undergone the same polishing process. In Figure 14, EBSD scans of the unheated and the heated sample are shown and the difference in crystal structure can clearly be seen.
Figure 14: EBSD scan of a sample that has been heated and one that has been heated with the dedicated heating stage.
Where the grains on the left have a diameter of a few μm, the grains in (b) have significantly grown to diameters of tens of μm.
In Figure 14a, very fine grains of a few μm in diameter are observed while in Figure 14b, the grain structure is much more coarse. Since both samples have the same history and the grains of sample (b) are significantly larger than those of its unheated counterpart (a), it is concluded that significant
(a) (b)
Page 23 grain growth has occurred during the heating of sample (b). The combination of the duration of the heating, which was in the range of tens of minutes, and the change in microstructure leads to the conclusion that the temperature has reached values well above 1200°C, i.e. the recrystallization temperature of tungsten.
This conclusion is supported by the fact that the SiC disk had apparently reached the limit of its operational temperature of 1600°C. Where the SiC disk had been grinded to a flat surface to create optimal contact sample and disk, upon disassembly of the stage it was found that the surface was no longer flat but showed significant surface change, leading to the suspicion of extremely high temperatures. In addition, various components had a black layer deposited on them, which is possibly caused by vapour deposition of SiC. Finally, on the backside of the tungsten sample, a yellowish layer was found. This is suspected to be tungsten trioxide, a material that forms when tungsten oxidizes at temperature of above 1500°C . By verifying if the formed material is indeed tungsten trioxide, a conclusion on the attained temperature can be drawn.
Unfortunately, besides the heating stage components, also the optic fibre ferrule was found to have a black deposition on its surface. This deposition has also taken place inside the ferrule, which is a steel cylinder with a hole along its centre that is approximately 100μm in diameter. Through this hole the laser light coming out of the fibre is guided to its goal and therefore, the hole must be clean. As some particle have found their way into the ferrule, the light path has been obstructed causing the light beam to shoot back into itself. This created several hotspots where the light burned through the different layers of coating rendering the optical fibre useless. As the optic fibre is coupled to the laser diode, the laser system must be repaired before any new experiments can be conducted.
The use of a thermocouple of 0.2mm in diameter has shown to be inadequate as the temperature shown by the thermocouple was around 950°C when the system became too hot. Where the hole in one of the alumina blocks has been designed to accommodate a thermocouple of 0.5mm in diameter, the delay of this thermocouple wire’s delivery led to the use of the thinner thermocouple.
As the thin thermocouple was not correctly fitted in the block, the temperature read-outs have shown to be too low. This led to the belief that the sample and SiC disk were at significantly lower temperatures than they were in reality. In addition, according to thermal simulations, operating at 40W should not have led to such elevated temperatures. However, the assumption of perfect contact between sample and SiC disk may have led to a larger heat flow away from the SiC in the simulations than occurred in reality. This means that in reality, the SiC disk has been more isolated than in the simulations, explaining a much higher temperature in reality than in the simulations.
While the heating assembly has become too hot, during the operation of the system before the fatal heating, it has been found that the SE image quality did not deteriorate and that the sample was stable. Upon heating, various points at the edge of the alumina hood have been observed to leave the field of view in a near to radial direction. This validates the design of the nest of springs as the thermal centre is located inside the field of view. As no images have been taken during the experiments, the previously mentioned sequence for the determination of the thermal centre has not been performed.
In addition to this thermal stability, no component in the SEM heated up to temperatures that may damage them. Also, the frame did not heat up during the operation of the heating stage. This leads to the conclusion that the cooling circuit and heatshield work as intended. In addition, as the imaging
Page 24 was stable, the use of laser heating over conventional heaters has been justified. Also, the use of the proposed simple vacuum feedthrough showed to work adequately resulting in vacuum levels comparable to that of conventional flanges. Together, these results lead to the conclusion that the majority of the dedicated heating stage and the supporting systems work as they were intended to.