Thermal protection of the SEM and the tensile stage

Similarly to the way in which the SEM is protected from the dedicated heating stage, the same principles can be applied for the tensile heating module. By limiting thermal conductance through narrow parts, minimizing contact area and using materials with low thermal conductivity, both the amount of hot mass and the energy needed to heat up this mass are kept to a minimum. In addition to the minimization of thermal energy in the system, the SEM can be protected from the system by using a heat shield. This heat shield can be designed in a similar fashion as that of the dedicated heating stage and could be modified from a hole to a slit if wider scan areas would be preferred over a small square.

Besides the SEM, the tensile stage must also be protected from thermal energy for it not to heat up.

This can be achieved by increasing the thermal resistance at the contact point between sample and stage, for example in the clamps. Additionally, besides the sample, all hot parts should be actively cooled by circulating water.

2.3 Minimization of thermal effects on data acquisition

As the factors that are detrimental to the image quality and acquisition remain the same as for the dedicated heating stage: thermal radiation, thermal expansion, and thermal drift, similar solutions must be found. Due to the dynamic nature of the tensile experiment, these solutions must differ from those created for the dedicated heating stage.

As previously mentioned, the sample cannot be pressed against a hood as it deforms during the experiment. This holds that, in the tensile case, the sample’s radiation may not only radiate freely from its top surface, but also from its sides and, depending on the heating system, also from its lower surface. To solve this problem, the heat shield must also shield the sides and possibly the bottom.

Contrary to the shielding, the accommodation of thermal expansion of the sample, is more easily achieved in the tensile case. In force controlled experiments and displacement controlled experiments, the thermal expansion in the length direction of the sample will simply result in, respectively, a larger displacement and a lower force. Lateral and vertical expansion aren’t a problem either as the expanding sample has no constraint due to its double clamped nature. The only precaution needed, is to minimize the amount of thermal displacement away from the scan area, which depends on the location of the thermal centre. Assuming that the scan area lies in the middle of the sample, the displacement is minimal if the sample’s thermal centre also lies in the middle of the tensile stage. If, for example the sample would be supported on one of the its sides, the lateral displacement would be significantly larger. This case is schematically shown in Figure 16 below.

Where the blue lines depict the axis around which the sample laterally expands (in the middle of sample (a) and the left edge of sample (b)) the cross sections A-A and B-B are used to depict the lateral expansion of the sample. As all points on the red line in the middle of the sample move in the same way, the analogy of supported beam from the dedicated heating stage also applies in this situation. Here, the red dot in the middle of the beam, may be perceived as the cross section of the aforementioned red lines. Thus the ratio of thermal displacement of any point on the red lines is

Page 32 given by l/(L0 + l). For a typical area of measurement of 100μm in the lateral direction and a sample of 1000μm in width, the thermal drift at 1500°C is reduced from 3.38μm, which can significantly interfere with EBSD maps, to an acceptable value of 338nm.

Figure 16:Schematic representation of the thermal expansion for both support cases: (a) the thermal centreline (blue) coincides with the sample middle, (b) the thermal centreline (blue) lies on the left edge of the sample. For both cases, the

beam analogy is shown on the left, where the red dot depicts any point on the sample centreline. The red crosses depict the location of points that were initially the edge of the field of view.

2.4 Coarse Design

As the design must be retrofittable to an existing tensile stage, the design space is limited and the heating module must consist of various part in order to fit in the tensile stage. The module consists of two hooks that can be clamped in the tensile stage and thermally isolate the stage from the heated sample. From behind, the heating system is inserted into the tensile stage. This system presses a heated rod against the back of the sample in order to heat the sample.

2.4.1 Design Components

As a large contact area between the sample and a clamp is equivalent to having a large potential for energy conduction, the choice is made to create a hook instead of a clamp. This hook can simply be clamped in the tensile stage and the sample hooked in. This does not only decrease the potential for energy dissipation, but also allows for the thermal isolation of the sample with respect to the tensile stage by inserting a thermal resistor between the sample and hook. In addition, as the tensile heating module is an additional, intermediate component, the system can be actively water-cooled without having to modify the tensile stage itself. The proposed assembly of hooks and sample is shown in Figure 17.

Page 33

Figure 17: Both hooks that can be clamped in the tensile stage. The holes in the sides are used to cool the hooks while the white inserts are zirconia inserts to isolate the sample (which is depicted in turquoise) from the hooks.

In order to isolate the sample, the hooks are manufactured out of tantalum for its low thermal conductivity. To lessen the heat flux towards the tensile stage, a badly conducting ceramic insert is used between the sample and the hook. In the dedicated heating stage, alumina was chosen as a thermal isolator in proximity to the field of view for its high maximum operational temperature and thermal conductivity. However, as the sample only heats up to 1500°C in the middle and remains below 1000°C at the edges, zirconia can be used as inserts at the hooks. Zirconia conducts even less than alumina and therefore isolates the tantalum hooks from the sample even better. Finally, water ducts are created in the body of the hooks to accommodate for active cooling and limit the increase in temperature of the stage.

Figure 18: (a) Cross section of the heating assembly with the heatshield where the heated rod and SiC disk are suspended from a plate, which is depicted in grey. Also, the inside of the cooling of the heatshield is shown (b) Close-up of the aforementioned suspension where the vertical rods are connected to the plate to provide the suspension. The black disk is

the SiC with the heated rod inserted on top (right)

Similarly to the technique used in the dedicated heating stage, for the tensile heating module, the system makes use of a SiC disk upon which the laser shines and heats up making the heating process independent of the sample material. However, as the SiC must be pressed against the sample for the heat to travel to the sample and the SiC must be shielded for the laser not to pass by it, pressing the SiC directly against the sample is difficult to accomplish. Therefore a system is made where the heating assembly is shielded from the tensile stage and the rest of the SEM, and only a small heated rod sticks out which is pressed against the sample. This is depicted is Figure 18 on the right. Here, the SiC disk is supported by a suspended platform. The platform is suspended from a thin sheet of steel

Page 34 that is connected to the cooling circuit. Steel is chosen as the local temperature in the plate may rise significantly. As the heating assembly is actively water cooled and the SiC is shielded, the tensile stage and SEM are under no influence of either light or heat, other than emanating from the sample.

Finally, to shield the chamber and components in the chamber form thermal radiation, a heat shield is placed over and around (to shield the tensile stage) the sample. Where for the dedicated heating stage, the heatshield had a conical hole machined into it, the heatshield for the tensile heating module has this same hole only it is also machined along a line to allow for a broader scan area. For the electron beam, the narrow hole is changed into a narrow slit. This heatshield, including the hooks and sample, is shown in Figure 19 below.

Figure 19: The heatshield containing the heating assembly (brown) including the hooks (grey) and the sample (turquoise) placed inside the tensile stage. The hole in the middle of the stage is where the electrons exit the material, and the slit left

to it is where the beam enters . The heatshield can be inserted from the back of the tensile stage and protects everything around the sample and hot hooks.

2.4.2 Thermal Simulations

To verify whether this coarse design meets the formulated design requirements, thermal simulations are performed. In addition, these simulations can be used to improve the design when the requirements are not yet met. To construct a valid and workable model, several assumptions and simplifications have been made. As the operation of the tensile heating module will occur under similar constraints as the dedicated heating stage, the majority of choices and assumptions in the numerical simulations, remains the same. First, the laser is modelled to be a simple heat flux incident on the backside of the SiC-disk for which the chosen value corresponds to the same energy put in to the system. As the experiments are conducted in vacuum, convection to the environment is neglected and black body radiation is assumed, i.e. this all surfaces that can ‘see’ each other are coupled via radiation. The only location convection occurs is in the cooling circuit of the heatshield,

Page 35 for which the convective heat transfer coefficient is chosen to be 500W/m²K [32]. All components that are in contact with each other are modelled to be in perfect contact. Also, the small bolts that hold various parts in place are neglected. Finally, the assumption is made that the heating module barely interacts with the tensile stage as the model is refined to the point that the hooks remain at a temperature close to the initial temperature. As there is a minimal temperature difference, no large heat fluxes will occur, through which heat exchange is deemed negligible. These choices are further motivated in Appendix E. Thermal simulations

In the simulations, a maximum power of 100W is used taking into account the maximum power of the laser. With this power, the temperature distribution of Figure 20 is found.

Figure 20: Temperature distribution of the coarse design for 100W of laser input. The sample temperature becomes 1100°C while the hooks reach a temperature of 290°C where they would be clamped in the tensile stage. The heatshield around the hooks and sample remain quite cold at around 40°C. The maximum temperature of 1400°C in the image is reached at the

back of the SiC disk and is therefore not visible in the image.

From this simulation, it is found that the cooling system below, and the cooling of the heatshield remove enough energy to maintain a low temperature of around 40°C. The sample temperature does, however, not reach the desired value of 1500°C but only reaches 1100°C. The fact that the sample does not heat up enough has two reasons. First, the thermal conductivity between sample and hooks is too large and therefore the heat flux away from the sample is too large to reach the goal temperature. This does not only result in a sample temperature that does not reach the goal temperature, but also in too high of a temperature at the extremities of the hooks that are clamped in the tensile stage.

In Figure 21, the assembly of sample and hooks is shown. Here, the sample reaches a maximum temperature of 1100°C while the extremities of the clamps reach 290°C. As these hooks are in direct contact with the tensile stage, the temperature of the hooks exceeds the maximum as the clamps of the heating stage would start to heat up, which is detrimental to the system and data quality. In addition, as the conducts in de hooks reach 317°C and the system is actively water cooled, this would heat the cooling system of the hooks too much for it to function properly as the water inside is can start to boil.

Page 36

Figure 21: Temperature distribution in the clamps with zirconia inserts (fine mesh on the hooks near the sample) for 100W of laser input where the sample reaches a temperature of 1100°C while the hooks reach a temperature of 290°C where

they would be clamped in the tensile stage.

Second, investigating the heating assembly leads to the conclusion that too much energy is allowed to flow away from the SiC disk as can be seen below in Figure 22. While the cooling circuit and top plate remain at a low temperature, the support for the SiC disk heats up considerably up to 1100°C.

This leads to the conclusion that the support allows for too much energy leakage from the sample to the cooling circuit and must therefore be revisited.

Figure 22: Heating assembly for an input power of 100W (left), top plate with the heated rod (top right), and the SiC disk with its support and the heated rod, where the pawls are suspended from the plate and use the plate as a spring to preload

the heated rod against the sample.(bottom right)

2.5 Optimized Design

Since the thermal simulations on the coarse design show that the clamps become too hot while the sample does not reach the goal temperature, some steps have been taken to create a design that overcomes these inadequacies of the coarse design.

Page 37 As the design space where modifications may be done has not changed, the exterior of the module remains the same. In addition, as the heatshield does an adequate job, it also remains the same as in the coarse design. The two major parts where changes have occurred are the hooks and the heating system. Where the hooks have been modified to meet the heating and thermal stability requirements, the heating system has mainly been modified for manufacturability reasons.

As the hooks did not thermally isolated the sample well enough in the coarse, the design has been revisited. Instead of four conducting legs, two V-shapes have been machined into the clamps. In these V-shapes, a zirconia insert is placed that isolates the clamps from the sample. This is done by reducing the contact area between the clamps and the inserts, but also be increasing the thickness of the inserts. The design of the new hooks is found below in Figure 23.

Figure 23: Modified clamps with the sample with the tungsten spring over the samples surface on the left. On the right, the sample is left out to show the zirconia insert, which is manufactured in one piece so that it supports the sample and is

easier to handle.

In addition to isolating the sample from the clamps, the combination the zirconia supports and the V-groove results in the thermal centre of the sample remaining stationary upon thermal expansion. The limitation of vertical movement is achieved by putting a tungsten spring over the sample at the extremities of the clamp. This has a the advantage of being negligible as a thermal influence due to a very small contact area and a small mass, and, with respect to the coarse design the optimized design also allows for simpler sample installation.

Figure 24: Temperature distribution for a sample temperature of 1540°C attained with a laser power of 65W where the cooling conduits only reach 75°C.

Page 38 Simulating the optimized design for a power of 65W gives the temperature distribution shown in Figure 24 above. Here, for a sample temperature of 1540°C, the maximum temperature at the extremities is just above 60°C and the temperature in the cooling ducts is at 75°C. While these temperatures are relatively high, they are within the system boundaries. In addition, the zirconia isolators do not become hotter than 1010°C which is well below their limiting temperature.

From the thermal simulation performed on the course design, is was found that the temperature was not large enough at the tip of the heated rod. For this reason the rod has been shortened. Besides the shortening of the heating connection, the suspended support for the SiC disk proved to be difficult to realise both in manufacturing as in the fulfilment of thermal conductivity. Therefore, the design of the heating connection has been modified, as can be seen in Figure 25.

Figure 25: Heating system of the optimized design. The cooling and heatshield haven’t changed while the suspension of the SiC-disk and heated rod have been changed to a tube around which a tantalum spring is located which presses the SiC disk

and heated rod against the sample.

Instead of suspending the support, the SiC disk and heated rod are now directly pressed against the rear of the sample. This is done by attaching a tantalum spring to the back of the SiC disk. The spring is centred around a tube that also serves as a light guide for the laser beam. This way, the spring and SiC disk remain in the correct position while the optical fibre is not in the vicinity of the hot disk. The tube is fixed to a plate that is bolted against the rear of the heating assembly. As the tantalum spring does not conduct heat well, the dissipated energy can easily be conducted into the cooling circuit to which the plate is connected.

In Figure 26, the temperature distribution around the heated rod and SiC disk is found. As the spring holding the disk in place is a very thin and long tantalum wire, it is not included in this simulation.

While the disk and rod attain a temperature of around 1650°C, the shielding around it reaches 450°C at its maximum and the temperature at the cooling circuit is 25°C. For the tube that centres the spring and guides the laser light, the maximum temperature, closest to the sample, reaches 300°C while the temperature at the bottom of the tube has a value of 40°C.

Page 39

Figure 26: Cross section of the heating assembly without the heatshield. The plate of the heatshield only reaches a temperature of 450°C while the top of the heated rod reaches a temperature of 1650°C.

As the heatshield has shown to work in an adequate fashion for the course design, no further simulations have been performed with the optimized design, since the new design puts a smaller heat load on the heatshield than the previous design. All the detailed drawings for the separate parts of the heating module for the tensile stage can be found in Appendix G. Detailed Drawings

As the heatshield has shown to work in an adequate fashion for the course design, no further simulations have been performed with the optimized design, since the new design puts a smaller heat load on the heatshield than the previous design. All the detailed drawings for the separate parts of the heating module for the tensile stage can be found in Appendix G. Detailed Drawings