Investigation of the electron beam sensitivity of mica for electron microscopy characterization

(1)

Electron beam sensitivity of mica

Investigation of the electron beam sensitivity of mica for electron microscopy characterization

Tess van Berkum S2940086 December 3, 2018

Daily Supervisor: Dr. J. Momand Supervisor:

Prof. dr. ir. B.J. Kooi

Supervisor:

Prof. dr. G. Palasantzas Faculty of Science and Engineering

Nanostructured Materials and Interfaces Zernike Institute for Advanced Materials

University of Groningen

Abstract

Mica is a mineral that has a naturally flat surface with basal and outstanding cleavage, which makes it a good substrate for high quality film growth.

In order to study the structure of these thin layers different geometries can be examined by (scanning) transmission electron microscopy ((S)TEM), namely plan-view and cross-section geometries. In this paper a dual beam system is used for preparing TEM-lamellas. The goal of the experiment is to investigate the electron beam sensitivity of mica when it is viewed in cross-section by a (S)TEM. It was found that mica has a certain electron dose threshold in order to conclude if it is damaged or not. The threshold lies in the electron dose range of 4.1∗10⁻⁷pC/µm²and 8.6∗10⁻⁷pC/µm² by 30 kV electrons. Furthermore an observation was made that the diffraction patterns vanish inhomogeneously, the out-plane lines decrease earlier than the in-plane lines when the electron dose increases. An equation of a sigmoidal logistic function, y = A2 + (A1 − A2)/(1 + (x./x0)^p), is found to express the vanishing of in-plane diffraction lines when the electron dose increases. The benefit of controlling the electron beam damage of mica is that the study of thin layers can be improved.

(2)

1 Introduction & Literature research

1.1 The growth of highly ordered thin films

The growth of highly ordered thin films is of great importance for the devel- opment and improvement of electron devices, such as applications related to electronics, optoelectronic, spintronics and thermoelectrics [1, 2, 3, 4]. Het- eroepitaxial growth is the growth of a crystalline film on a crystalline substrate with a different material. Usually substrates with smallest lattice mismatch are chosen for the heteroepitaxial growth of highly ordered thin films. This is because the epitaxy is extremely influenced regarding film morphology and crystallinity which can even depend on the termination of the single atomic layer surface of the substrate. For example the epitaxial film growth of Sb2Te3 and GeTe on Si(111) shows that randomly twisted domains can occur [5]. Mica is often used as a substrate for the study of thin films, because of its flat surface.

When for example polyethylene is deposited on a mica surface, the (001) plane of the polyethylene crystal film can be parallel to the (001) surface of the mica.

Moreover the polyethylene crystal film grows according to the mica (001) surface structure, this means that almost no lattice mismatch occurs [1]. One promising method to improve the crystalline quality is to grow thin films on weakly inter- acting substrates via van der Waals epitaxy (vdWE), which assist the growth of a defect-free epilayer with its own lattice specifications from the first layer.

Thin multilayer films can also grow on substrates, for instance to investigate the strain state of the individual layers and so optimize the functional performance of vdWaals heterostructures. The images of the trilayers Bi2Te3–Sb2Te3–Bi2Te3

grown on Si02substrates show a perfect out-of-plane texture which means that the layers are of high quality [6]. Another example is the continuous films growth of continuous ReS2films on mica substrate. The multilayer films shows a layer-by-layer growth mode with strong (001) orientation [7]. Muscovite mica is a generally known substrate that prefers the vdWE growth of semiconductors, metal thin films, and oriented nanowire arrays. Mica’s excellent electrical, thermal insulation, high transparency in ultrathin thickness, and flexibility makes it an optimal substrate for thin film growth and the investigation of the corresponding electrical and optical properties. High-quality Bi2Te3 thin films have been made on insulating and flexible mica substrates, presenting atomically smooth plateaus of a few microns. This means that the naturally flat surface of mica makes it a suitable substrate material for the study of thin films [2].

1.1.1 Methods to study thin films

In order to get a thorough study of thin films using (S)TEM, first a TEM sample has to be produced. Different techniques are used for making TEM samples, one of them is a water-transfer technique. For instance a film is grown on a mica substrate and separated in a water bath, in which the active layer simply floated off the mica substrate. After the float off, the film can be picked by a TEM grid. In this technique the mica is only used as growth template, because of the mica’s flat surface. [3, 8]. Another technique for making TEM samples is cutting out a lamella from the sample and mount it on a TEM grid, in this way the cross section of the sample can be analyzed. This technique is used in

(4)

the paper, see section 2.1 for a detailed description. One can analyze plan-view and cross section geometries or plan-view imaging and plan-view selected area electron diffraction (SEAD) of the sample, see figure 1. Generally the water- transfer technique is performed with a mica substrate to analyze the thin film plan view and plan view SAED. Analyzing the cross section geometries is mostly done for thin film grown on silicon substrates, as can be seen in figure 1. In this report we make use of a cross section analyzes from a sample of a thin film grown on a mica substrate. the cross section analyzes can demonstrate the

Figure 1: An illustration of the top-view, cross section and plan-view SEAD of a sample, received from J. Momand.

high quality of the film and the atomically distinct interfaces between substrate and thin layers [5]. Plan view SAED, is a technique that can be performed inside a TEM, a TEM sample is illuminated by a parallel beam of high-energy electrons. The electrons are treated as wave-like electrons, because the spacing between the atoms in the sample is approximately a hundred time larger that the wavelength of the electrons. Electron diffraction comes from the combined scattering from many atoms arranged in a crystal. Some part of the electrons will be scattered to specific angles, determined by the crystal structure of the sample, while other electrons pass through the sample without scattering. Therefore, a number of spots will be seen as image, called the SAED pattern. Each spot corresponds to a diffraction condition of the sample’s structure. SAED patterns are a projection of the reciprocal lattice, diffraction spots can be seen as lattice reflections. Furthermore, SAED patterns can be used to determine crystal structures, analyze crystal defect and measure lattice parameters by tilting a sample to low-index zone axes. SAED patterns can look like ring patterns if they are taken from crystals with random orientation (polycrystalline) and spot patterns if they are taken from a single crystal. To obtain the diffraction pattern of a selected area, a selected area aperture is inserted into the beam path below the sample holder to block the beam. Different sized apertures can be chosen. SAED is comparable to X-ray diffraction, however with SAED, areas of small dimensions of hundred nanometers can be examined. In TEM, plan-view, cross-section view and SAED of TEM samples can be performed [1, 6, 9].

(5)

1.2 Muscovite mica

Mica is a group of diverse coloured or transparent mineral silicates which exist by nature and without artificial aid, it can be acquired in large numbers. It has a high strength, good thermal stability and good corrosion resistance. The crystal structure is monoclinic and the shape can be described as hexagonal with basal and outstanding cleavage. The easy cleavage is along the (001) plane, see figure 2 [10]. The surface of muscovite mica can be described as flatter than a pancake [11]. The latter gives mica its unique property, because it can be simply divided into small parts while remaining a relatively high strength thanks to the other two axes [12]. Muscovite mica is the kind of mica that is used in this

Figure 2: Muscovite mica structure illustrated in a ball and stick model, showing the easy cleavage along the (001) plane and a glide plane along the c-axis [11].

paper. It is a potassium-rich mica with the following generalized composition KAl2(AlSi3O10)(OH)2. In this formula potassium is sometimes replaced by other ions with a single positive charge such as sodium, rubidium, or cesium.

Aluminum is sometimes replaced by magnesium, iron, lithium, chromium, or vanadium. Muscovite mica is used in a variety of products, as in joint paint, plastics and rubber. Sheet mica has several properties that makes it suitable for very special uses, namely it can be split into thin sheets and the sheets are chemically inert, dielectric, elastic, insulating and refractive. Moreover it is stable when exposed to electricity, light and extreme temperature. Most sheet mica is used to make electronic devices [13]. Muscovite mica is the best of all micas in dielectric strength, perfection of cleavage and transparency [14].

1.3 Radiation damage and mica amorphization

The interaction between the electron beam and the sample in an electron microscope, causes radiation damage. Radiation damage is the formation of structural or compositional changes caused by the electron beam. In fact, radiation damage is the loss of crystallinity or loss of mass caused by the electron beam. This damage restricts the extent of information that can be collected from an electron

(6)

microscope sample [15]. Moreover a gas ion beam can cause radiation damage.

The microstructure of muscovite mica exposed to a rare gas ion beam has been studied by TEM, the investigation of damage was carried out using argon and helium ions of enough energy to go over the 100–150 nm mica samples. Through the complete disappearance of the diffraction maxima from crystalline mica the amorphization fluence was determined. The amorphization fluence is defined as that required for the full amorphization of muscovite mica. [16]. As mentioned in section 1.1.1 the diffraction pattern gives information about the crystallinity of the material and hence about the amorphization of the material.

1.4 The goal of the experiment and equations

The goal of the experiment is to investigate the electron beam sensitivity of mica when it is viewed in cross-section by a (S)TEM. Gaining some experience in working with TEM and a dual beam system to make and analyze TEM- samples is an integral part of the research. When looking at a cross section of a thin film sample, the thin layer(s) and the substrate are viewed. In this experiment a sample of thin layers of bismunth telluride (Be2Te3) and tungsten ditelluride (WTe2) grown on mica is used. In figure 3 it can be seen that mica is easily affected when it is looked at by a TEM. In figure 3b, in the dark spot, the atomic structure of mica can not be seen anymore. Furthermore a crack can be seen, which indicates bending of the material. Moreover the structure of the thin layer close to the mica surface can not clearly be studied. This means that if the

(a) (b)

Figure 3: (a) cross-section TEM picture of a lamella with mica substrate and a thin film, (b) a zoom in of the left picture, received from J. Momand.

beam damage on the substrate mica can be controlled, the study of thin layers will be improved. That is why my report is about the electron beam sensitivity of mica when it is viewed in cross-section by a (S)TEM. To investigate if mica can be imaged by a (S)TEM without damage, the electron dose will be adjusted by changing the current at a certain area of the TEM sample by a dual beam system in STEM mode. For obtaining the applied electron dose, the machine’s electron dose value is taken. For calculating the electron dose needed to take

(7)

an image of the sample area, the following formula is used,

D = I ∗ t

A (1)

with electron dose D (pC/µm²), current I (pA), time t (s) and area A of sample image (µm²). Analyzing the diffraction patterns of the radiated mica will give inside in the electron beam sensitivity of mica, the amorphization of mica. The sample related contribution to the diffraction peak profiles is often described as a combination of Gauss and Lorentz (Cauchy) functions [17]. In this report a Lorentzian function is the better fit through the central peak intensity data points. The Lorentzian function can be fitted by using the following formula,

L(x) = y0 + Aw

4(x − xc)²+ w² (2)

with L(x) the Lorentzian function, y0 the start value of y, A a factor, w the width of the central peak and xc the centering x value of the peak. For calculating the area below the central peak, the line integral of the lorentzian fit Iarea(x)is obtained by the following equation,

I_area(x) = Z b

a

Aw

4x²+ w²dx (3)

with a and b the start and end value of x of the central peak. The error in w and A, σw and σAcan be found in Matlab, see Appendix 7.5, and implemented in the following formula to acquire the error in the integral σI_area(x),

σ_I_area_(x)= r

(arctan(100

w ) ∗ σA)²+ (− 100A

10000 + w² ∗ σw)². (4)

2 Research strategy and technique used

2.1 Lamella

For preparing TEM samples, the Helios G4 CX DualBeam for material science (Helios) is used. The Helios is a Thermo Scientific DualBeam system, where a focused ion beam (FIB) is combined with a scanning electron microscope (SEM). One of the most important function of a FIB workstation is preparing samples for TEM investigation, because the samples must be uniformly thin to get the analyzing beam of electrons through the sample and create an image.

The FIB provides not only the preparation of large, uniformly thick, section specific samples, in fact it can also make lamella used for TEM samples from combined samples consisting of inorganic and organic materials with very different properties. No other technique can select the target area as precisely as a FIB can, with care a lamella can be prepared within a spatial accuracy of 20 nm. Hence FIBs are widely used to produce TEM cross-section lamella [18]. In order to prepare TEM cross-section sample lamella a thin piece of material is cut from the area of interest in the sample. Before cutting, a platinum layer is deposited on this area to protect the top portion of the specimen from the gallium particles of the ion beam and to mark the position of the lamella, figure 4a. Trenches are then milled on both sides of the platinum layer to create a 1

(8)

µm thick lamella, figure 4b. Using a needle manipulator glued to the sample by applying platinum, figure 4c, to perform the sample lift-out. The lamella is attached to a Cu TEM half-grid, figure 4d and a final FIB thinning and cleaning are performed in order to obtain a lamella of about 200 nm thick [19, 20]. For more extended information about making TEM-lamellas for this experiment, see Appendix 7.1. In some figures, for example figure 4b, flashes of light can be seen due to charging effects of mica from the ion and electron beam. For

(a) Deposition of platinum. (b) Milling of both sides.

(c) Needle is glued to the lamella. (d) Lamella attached on Cu TEM half-grid.

Figure 4: The process of making a TEM-lamella.

studying the lamella in the (S)TEM, it holds that for a thinner lamella more electrons can travel through the material, which means that more electrons can be detected and that will give a clearer signal. For this experiment the thin layers won’t be investigated but the substrate itself, an around 200 nm thick homogeneous mica lamella was the thinnest possible lamella to make without significant bending of the material. The 200 nm thick lamella still gave a clear signal in the STEM, see figure 5a. Moreover the upper half of the lamella has almost non colour change which indicates that the upper half of the lamella is quite uniformly thin. As you can see in figure 5a, the thin film on the mica substrate is almost entirely gone, except for a small piece, this is due to the thinning and cleaning part of the lamella. Moreover some platinum can be seen, which indicates the place where the needle was attached to perform a lift out of the lamella.

(9)

(a) STEM image of the lamella.

(b) STEM image of the lamella with boxes.

Figure 5: STEM images.

2.2 30 kV STEM experiment

To investigate the electron beam sensitivity of mica the following experiment is performed on an approximately 200 nm homogeneous mica lamella. The Helios was put in STEM mode and the lamella gave a clear signal for imaging at 30 kV electrons, see figure 5a. An array of squares of dimensions 0.6 µm by 0.6 µm were illuminated by electrons at the upper half of the lamella for 60 seconds at 30 kV with different current. The most left square, called box 1, was illuminated with an electron current of 22 nA, After box 1, box 2 is illuminated with a current of 11 nA, box 3 with a current of 5.5 nA, box 4 with a current of 2.8 nA, box 5 with a current of 1.4 nA and box 6 with a current of 0.69 nA,

(10)

see figure 5b and table 1. The boxes were created one at a time, because the charging effect of the mica would be minimized and hence the drifting of the boxes. The machine determines the applied electron dose and is defined as the electron dose per pixel. For instance, for illuminating box 1 with electrons, each pixel is 0.02 µm by 0.02 µm, the loop time is 758 µs and 79126 passes are done, the corresponding electron dose is 3.4∗10⁻⁶pC/µm². The Helios automatically adapts these values for every box.

2.3 200 kV TEM experiment

The diffraction pattern of the boxes will be analyzed in order to say if the mica is damaged in the boxes. These patterns can show if the crystallined mica is transformed into an amorphous one. To obtain the diffraction patterns the lamella with the boxes is placed in the TEM. An overview image with magnitude 12k, 200 kV and a current density of 5.2 pA/cm² is made, figure 6a. When making use of the experimental SAED technique, the diffraction pattern inside the boxes can be obtained. For this the 3rd aperture is chosen, because this aperture was small enough to fit inside the boxes and the diffracted signal was still high enough. The circles in figure 6b indicate at which place a diffraction pattern was obtained. So for each diffraction pattern in the box also a background diffraction pattern was taken. As can be seen in figure 6 only box 1-5 are visible on the lamella, that is why only the diffraction patterns of these boxes are obtained.

(a) TEM image of the lamella. (b) TEM image, diffraction spots indicated.

Figure 6: TEM images.

When making images or diffraction patterns of the lamella it is very important that this is done with a low current for not damaging the material any further.

For a view to see what happened to the mica in the boxes, real space images of the damaged boxes in magnification 500 and 800 were taken. Real space images of the boxes were obtained with a higher electron dose than the diffraction patterns, that is why the diffraction patterns were obtained first.

2.4 Analyzing diffraction patterns

The software programs DigitalMicrograph and Matlab are used to analyze the obtained diffraction patterns. For the obtained diffraction patterns it holds that

(11)

the highest intensity spot is called the central spot. The following is analyzed:

1. The intensity difference in the central peaks per dose.

In order to get the data for the intensity of the central spot, the option Draw a line profile in DigitalMicrograph is used with an integration width of 20 and it is placed over the diffraction line with the central spot. This is done for all the boxes, the dimensions of the Line Profile stays the same. In Matlab the data can be plotted and by the use of equation 2 a Lorentzian fit can also be plotted through the data points. Furthermore the central peaks can be aligned to get a clear overview of the decreasing of the peaks per decreasing dose. With this data and equations 1,2,3,4, the peak width versus the dose and by means of the line integral, the area under the central peaks versus the dose can be plotted.

2. The decreasing of the in- and out-plane diffraction lines per increasing dose.

In order to get the data for the decreasing of the in- and out-plane diffraction lines per dose, the option Draw a line profile in DigitalMicrograph is used with an integration width of 20 and it is placed over all the diffraction lines. All peaks of intensity of 8 counts or more are counted per line, this is done for all boxes. Furthermore all the diffraction spots per line per box are counted with the naked eye. An error can be established by calculating the average difference of the number of peaks found by means of the line profile and by means of the eye. The line with the highest intensity peak is called the 0th order line, the line above and below the 0th order line is called the 1st order line, the 2nd line above and below the 0th order line is called the 2nd order line, etc. In Matlab the data of the average number of peaks per order line can be plotted.

2.5 Uniformly thin lamella

The 30 kV STEM experiment is performed at the upper half of the lamella, because it looked approximately uniformly thin. To get more inside on how uniformly thin the upper half of the lamella is, a line profile in DigitalMicrograph is obtained form the row of boxes and the row of backgrounds in figure 6b. The circles are taken as points to be within the line profile, hence an integration width of 30 is used. From the line profile the location of the circles in pixels could be noted. The average background intensity per background can be calculated and this can give an indication of how uniformly thin the lamella is. Hence deviations in intensity values can be adjusted if necessary.

3 Results

The obtained diffraction pattern and real space images for box 1, box 2 and box 3 are displayed in figure 7 and figure 8.

3.1 The intensity difference in the central peaks per dose

The difference of the central peak per dose is analyzed and by the use of equations 1,2,3 and 4 the doses, peak width, line integral with errors of the central peaks are calculated by the corresponding doses, see table 1. In figure 9 the

(12)

(a) Dose = 3.4 ∗ 10⁻⁶ pC/µm²

(b) Dose = 1.7 ∗ 10⁻⁶ pC/µm²

(c) Dose = 8.6 ∗ 10⁻⁷ pC/µm²

Figure 7: Diffraction patterns.

(a) Dose = 3.4 ∗ 10⁻⁶ pC/µm²

(b) Dose = 1.7 ∗ 10⁻⁶ pC/µm²

(c) Dose = 8.6 ∗ 10⁻⁷ pC/µm²

Figure 8: Real space images.

Boxes Current (nA) Dose ∗10⁻⁶(pC/µm²) IArea(x) ∗10³ σ_I_Area_(x)∗ 10² w (pixel) σw(pixel)

Box 1 22 3.4 17 3.5 4.9 0.12

Box 2 11 1.7 13 2.4 4.2 0.096

Box 3 5.5 0.86 3.3 0.78 3.5 0.099

Box 4 2.8 0.41 1.2 0.27 3.6 0.093

Box 5 1.4 0.22 0.78 0.17 3.4 0.093

Background 2 0 0 1.3 0.34 3.1 0.10

Table 1: Boxes with corresponding current and calculated data.

Lorentzian fit through the data points of the central peak can be observed. In figure 10 the line profiles of the 0th order line are aligned to get a clear overview.

Figure 9: Lorentzian fits through the central peaks data points

The central peak width versus the dose and the line integral, number of detected

(13)

Figure 10: Aligning of the central peaks.

electrons versus the dose is plotted in figure 11. The software Origin produced

(a) (b)

Figure 11: In the figures are plotted (a) central peak width versus the dose and (b) number of detected electrons versus the dose.

by OriginLab Corporations is used to get a fit through data points. A logistic function which is a common "S" shape is found to be the best function through the data points for fitting the peak width versus the electron dose within the error bounds,

y = A2 + (A1 − A2)/(1 + (x./x0)^p) (5) with A1 = 3.3, A2 = 5.1, x0 = 1.9∗10³and p = 2.7. The same function is found to be the best fit for the number of detected electrons versus the dose within the error bounds, namely equation 5, with A1 = 1.0 ∗ 10³, A2 = 1.7 ∗ 10⁴, x0 = 1.4 ∗ 10³ and p = 4.0.

3.2 The decreasing of the in- and out-plane diffraction lines per increasing dose.

In table 2 the data for the counted peaks per order can be found. The diffraction patterns weren’t completely taken in the zone as of the mica, that is way the 2nd order line is counted ones. The average error of 1.46 peaks is round off to the average error of 2 peaks, because peaks can only be counted as whole peaks.

(14)

Box 1 Box 2 Box 3 Box 4 Box 5 Background 2 Order Graph Eye Error Graph Eye Error Graph Eye Error Graph Eye Error Graph Eye Error Graph Eye Error

2 20 24 4 27 30 3 24 26 2 1 1 0 0 0 0 23 30 7

1 21 24 3 26 26 0 27 27 0 14 14 0 4 4 0 29 29 0

0 22 28 6 25 25 0 25 29 4 14 14 0 4 4 0 23 25 2

1 18 19 1 23 25 2 28 28 0 4 4 0 2 2 0 17 19 2

Table 2: Counted peaks per order per box.

In figure 12 the data of the average number of peaks per order line and the plotted fit can be seen. The software Origin is used to get a fit through these

Figure 12: Number of peaks per dose plotted.

data points. A sigmoidal logistic function, type 1 in Origin, which indicate also a "S" curve, is found to be the best function through the data points for fitting the number of peaks versus the dose within the error bounds,

y = a/(1 + exp(−k ∗ (x − xc))) (6)

with a = 27, xc = 2.1 ∗ 10³and k = −1.6 ∗ 10⁻³.

3.3 Uniformly thin lamella

The obtained values for the average background intensity per background can be seen in table 3. The difference as well as the percentage is obtained by using

Background Average intensity (counts) Difference (counts) Percentage (%)

1 3335 51.78 1.6

2 3284 0 0

3 3270 -13.56 -0.41

4 3309 25.26 0.77

5 3337 53.57 1.6

Table 3: Average background intensities.

background 2 as reference point.

(15)

4 Discussion

The central spots of the diffraction patterns aren’t in the middle of the images, because the electron beam in not fully placed in the zone axis of mica. Placing the electron beam in the zone as of mica is very difficult in this case, due to the locally applied deformation of the lamella, which causes a small kind bending of the lamella. Nevertheless the highest intensity spot is still called the central spot.

4.1 The process of making TEM-samples

It is difficult to get an uniformly thin lamella that is thin enough for TEM signal.

When making the lamella too thin the lamella will bend and that makes it useless for this experiment. Moreover in figure 5b cracks in the mica lamella can be seen, which indicates a small kind of bending of the lamella. The short time diffraction will probably not significantly damage the mica any further, because it is obtained with a very low electron current. However when an overview picture is taken the mica is in interaction with the electron beam, which means that when performing the experiment it has to be done efficient and quickly.

the picture of figure 3 is taken in a TEM at 200 kV, 20 pA at a sample area of approximately 15 nm by 15 nm for 10 µs. with the use of equation 1, the electron dose that was needed to take this picture with 1024 by 1024 pixels is 0.88 pC/µm², so the applied electron dose per pixel is approximately 8.8

∗10⁻⁷ pC//µm². When comparing this value with the graphs in figures 11, the conclusion can be made that by taking this picture the mica is damaged.

The lamella that is used in the experiment is not entirely uniformly thin, which means that there is some deviation in the intensity data points. As can be seen in table 3 the largest deviation of the average background intensity is less than 2 percent from background 2, this will give no significance difference in the other experimental data. That it why, the intensity difference is not taken into consideration for the rest of the experiment and the lamella is considered to be uniformly thin.

4.2 The intensity difference in the central peaks per dose

In figure 9 and figure 10 the intensity of the central peaks can be seen, the box with the highest doses has the highest intensity and so it decreases in dose and intensity. Equation 5 is a function of the dose and intensity, which indicates a logistic function. The curve indicates that there is a threshold for the damaged mica. From figure 11 can be seen that the threshold starts somewhere between dose 8.6 ∗ 10⁻⁷ pC/µm² and 4.1 ∗ 10⁻⁷ pC/µm² and especially from figure 11b can be said that the threshold is nearer to the 4.1 ∗ 10⁻⁷ pC/µm² data point.

The data point for background 2 is included in the graphs to know at which point the mica has the least damage, that is why the data of background 2 is taken as reference point.

(16)

4.3 The decreasing of the in- and out-plane diffraction lines per increasing dose

In figure 7a decreasing of in- and out-plane diffraction lines per increasing dose can be seen. Furthermore it seems that the lines decrease unequally. The out- plane lines decrease earlier than the in-plane lines do. Moreover this can be seen in the real space images of box 2 in figure 8, because the box is damaged unequally. The left and right sides of the square is less damaged than the top and bottom of the square. In figure 12 a sigmoidal logistic function is fitted through the data points with the help of Origin. For the 0th and 2nd order fit the program converted the equation easily, but for the 1st order the program didn’t converge. After fitting equation 6 in Matlab a warning was given by the 1st order fit. Warning: The Jacobian at the solution is ill-conditioned, and some model parameters may not be estimated well (they are not identifiable). Use caution in making predictions. That is why no further pronunciations can be done about the 1st order fit. The 2nd order line below the 0th order line is not counted for within the calculations for the obtained data points, because of the not fully placed electron beam in the zone axis of mica. As regards the 0th and 2nd order fit, it can be seen that the 2nd order fit drops down to zero much faster than the 0th order fit. Which means that the second order line vanishes must faster than the 0th order line when dose is increased. Again an electron dose threshold can be seen in between the doses of box 2 and box 3 for both order fits. No statement can be made about the doses step height for out-plane diffraction line to vanish, because of the inconclusive 1st order fit. Nevertheless equation 6 is a function of the vanishing of in-plane diffraction lines. Probably if more data points were available, a well conditioned fit could also be made through the 1st order data points.

Due to a time limit for the experiment it wasn’t possible to obtain more data points for presenting the above noted findings more confident. Especially more data points are needed around the doses of box 2 and box 3 for getting an exact value or range for the threshold. For improving the experiment more data points need to be obtained for the same range of dose.

5 Conclusions

The purpose of this experiment was to investigate the electron beam sensitivity of mica when it is viewed in cross-section by (S)TEM. Furthermore gaining some experience in working with TEM and a dual beam system to make TEM samples was an integral part of the research. TEM samples were created by cutting lamella out of a sample, consisting of thin layers of Be2Te3and WTe2grown on mica with means of a dual beam system. Afterwards the lamella was attached to a TEM grid. To investigate the electron beam sensitivity of mica, boxes of different doses were created on the lamella with the dual beam system in STEM mode. Diffraction patterns and real space images of these boxes were obtained by a TEM and analyzed by means of the software programs DigitalMicrograph, Matlab and Origin. The curves of figure 11 indicates that there is a threshold for the mica to be damaged for 30 kV electrons. This threshold is lies in the electron dose range of 4.1 ∗ 10⁻⁷ pC/µm² and 8.6 ∗ 10⁻⁷ pC/µm², probably

(17)

the threshold is closer to the value 4.1 ∗ 10⁻⁷ pC/µm². After this value can be concluded that the mica is damaged and before this value can be concluded that the mica is not significantly damaged. Furthermore an observation was made that the diffraction lines decrease unequally, the out-plane lines decrease earlier than the in-plane lines, equation 6 is found to express the vanishing of in-plane diffraction lines when the electron dose increases.

5.1 Outlook

As these thesis has shown, there is a certain threshold for the mica to be damaged. This opens a door to a lot of future experiments to gain as much as possible information on the electron beam sensitivity of mica, eventually for improving the study of thin films grown on mica substrates. Suggestions for further research could be to:

- Improve the experiment done in this report by obtaining more data points.

In this way a more exact value or range of the electron dose’s threshold by 30 kV electrons can be found.

- Perform approximately the same experiment as done in this report by only using the TEM. This is because thin layers are mostly studied in TEM for high resolution imaging. The kind of experiment that is possible in TEM is to illuminate a fixed area of the lamella with a definite low electron current and constant electron voltage for a specific time after that obtain a diffraction pattern and a real space image of this area. Illuminate the area again for the same amount of time, electron current and electron voltage and obtain a diffraction pattern and a real space image of this area, etc. In this way the dose is approximately summed up every time and the unequal thickness of the lamella is no issue anymore.

- Perform the above mentioned experiments with different electron voltages.

As can be seen in figure 17 in Appendix 7.4, a lamella with boxes is imaged with different electron voltage. When the electron voltage increases, more electrons travel through the sample and the signal gets clearer, hence a clearer image of the lamella. In the abovementioned experiments the voltage is a fixed value and to obtain more information about the electron sensitivity of mica with different electron voltages, a couple of the same experiments mentioned above can be performed with different electron voltages.

In this way a lot of data about the electron sensitivity of mica can be obtained.

(18)

6 Acknowledgement

I would first like to thank my supervisor Prof. dr. ir. B.J. Kooi of the Fac- ulty of Science and Engineering at the University of Groningen for extending information about possible bachelor research projects in his field of research and welcoming me to the research group Nanostructured Materials and Interfaces.

The door to prof. Kooi office was always open whenever I had a question or needed some advice. He encouraged me to make this research my own work.

I would also like to acknowledge my second supervisor Prof. dr. G. Palas- antzas of the Faculty of Science and Engineering at the University of Groningen as the second reader of this thesis and for welcoming me to the group.

I would also like to thank Gert ten Brink of the Faculty of Science and En- gineering at the University of Groningen for opening his office to me. What is more, to have a place to come to in the bachelor thesis period and hence for the opportunity to get insights about my bachelor thesis from colleagues, Sytze de Graaf and Paul Vermeulen over a cup of tea or coffee.

I would especially like to thank my daily supervisor Dr. J. Momand of the Faculty of Science and Engineering at the University of Groningen for shar- ing expertise and valuable guidance and encouragement extended to me. He made this bachelor research project possible in a limited time by guiding me through the experimental face due to his highly operating skills in a dual beam system and electron microscopes. Moreover he provided me with information and figures on the subject of matter and provided me with advise on how to write a report, feedback included. The door to J. Momand office was always open whenever I ran into a trouble spot or had a question about my research or writing. He steered me in the right direction whenever he thought I needed it.

Thank you.

Tess van Berkum

(19)

References

[1] H. J. Gao C. Zhu Z. Ma Z. Q. Xue, W. M. Liu and S. J. Pang. Polyethy- lene crystal lamella growth on the mica. Journal of Vacuum Science &

Technology, 10:627–629, jul 1992.

[2] Weiyi Wang N. Meyer L. H. Bao L. He M. R. Lang Z. G. Chen X. Y. Che K. Post J. Zou D. N. Basov K. L. Wang K. Wang, Yanwen Liu and Faxian Xiu. High-quality bi2te3 thin films grown on mica substrates for potential optoelectronic applications. Applied Physics Letters, 103, 2013.

[3] K. Kikuchi Y. Saito, N. Fujimoto and Y. Achiba. C76 thin films grown on mica and nacl substrates. PHYSICAL REVIEW B, 49(20), may 1994.

[4] Xiaobo Li Gang Wang Kaiqiang Liu Zhou Yang Qingliang Feng Xing Liang Zhongyue Zhang Shengzhong Liu Zhibin Lei Zonghuai Liu Hua Xu Fang- fang Cui, Cong Wang and Jin Zhang. Tellurium-assisted epitaxial growth of large-area, highly crystalline res2 atomic layers on mica substrate. Ad- vanced Materials, 28:5019–5024, 2016.

[5] Ruining Wang Raffaella Calarco Jamo Momand, Jos E. Boschker and Bart J. Kooi. Tailoring the epitaxy of sb2te3 and gete thin films using surface passivation†. CrystEngComm, 20:340–347, 2018.

[6] Jamo Momand Bart J. Kooi Paul A. Vermeulen, Jefta Mulder. Strain engineering of van der waals heterostructures†. Nanoscale, 10:1474–1480, 2018.

[7] Xiaobo Li Gang Wang Kaiqiang Liu Zhou Yang Qingliang Feng Xing Liang Zhongyue Zhang Shengzhong Liu Zhibin Lei Zonghuai Liu Hua Xu Fang- fang Cui, Cong Wang and Jin Zhang. van der waals epitaxy of large-area continuous res2 films on mica substrate. RSC Advances, 7:24188 –24194, 2017.

[8] A. K. Lemmens W. de Poel P. H. J. Kouwer A. E. Rowanac J. Feenstra, M.

van Eerden and J. J. Schermer. Muscovite mica as a growth template of pc61bm crystallites for organic photovoltaics†. CrystEngComm, 19:4424–

4436, 2017.

[9] Sven Hovmöller Xiaodong Zou and Peter Oleynikovang. Electron Crystal- lography: Electron Microscopy and Electron Diffraction. Oxford Scholarship Online, 2012.

[10] Zhuang Wu, Derek Li. Mica : Properties, Synthesis, and Applications.

Nova Science Publishers, Inc, 2012.

[11] Jakub Drnec Francesco Carla Roberto Felici Peter Mulder Johannes A.A.W. Elemans Willem J.P. van Enckevort Alan E. Rowan Elias Vlieg.

Wester de Poel, Stelian Pintea. Muscovite mica: Flatter than a pancake.

Surface Science, 619:19/24, 2014.

[12] I.J. Kemp D.M. Hepburn and A.J. Shields. Mica. IEEE Electrical Insula- tion Magazine, 16(5):19–24, sep 2000.

(20)

[13] RPG Hobart M. King, Ph.D. Muscovite.

https://geology.com/minerals/muscovite.shtml. Accessed on 05/10/2018.

[14] Electron Microscopy Sciences. Mica sheets and disks.

https://www.emsdiasum.com/microscopy/products/preparation/mica.aspx, 2018. Accessed on 19/09/2018.

[15] Yeshayahu Talmon. Electron Beam Radiation Damage to Organic and Bi- ological Cryospecimens. Cryotechniques in Biological Electron Microscopy, 1987.

[16] Argonne IL R Birtcher, Argonne National Laboratory (ANL). Ion beam amorphization of muscovite mica. Journal of Materials Research, 11(7), July 1996.

[17] Frank Girgsdies. Peak profile analysis in x-ray powder diffraction.

http://www.fhi-berlin.mpg.de/acnew/department/pages/teaching/pages/, teaching wintersemester 2015 2016/, frank girgsdies peak profile fitting in xrd 151106.pdf, Electron Microscopy Group, Department of Inorganic Chemistry, Fritz-Haber-Institut der MPG, Berlin, Germany. accessed on 16 november 2018.

[18] T. Kamino J. Mayer, L. A. Giannuzzi and J. Michael. Tem sample preparation and fib-induced damage. MRSBulletin, 32, may 2007.

[19] Hoi Pang Ng Dacian Tomus. In situ lift-out dedicated techniques using fib–sem system for tem specimen preparation. Elsevier, 44:115–119, 2013.

[20] FEI. Helios dual beam training, module 5 tem sample preparation. Man- ual, 5350 NE Dawson Creek Drive Hillsboro, Oregon 97124 USA, 7 2018.

confidential.

(21)

7 Appendix

7.1 TEM-lamella

The following operation procedures are executed in order to prepare TEM cross- section sample lamella.

- Protective layer deposition.

When starting the first procedure, the SEM is necessary to find the area of interest with a recognizable feature to set up eucentric height, figure 13a. When all settings are set the protective layer can be deposited by E beam and I beam platinum, figure 13b,c, by inserting the Pt GIS needle.

The E beam Pt protects the top surface for a time against the ion beam, mostly until the I beam pt is deposited.

- Bulk & intermediate mill.

After the deposition of the protective layer, the lamella needs to be shaped or milled from the bulk sample. This starts with bulk milling and intermediate milling to reveal the lamella by drawing a regular cross section on the top and bottom side of the Pt layer and mill, figure 13d. The lamella has to be ready for lift out, this is done by making an U shape cut. Only a small part of the right side of the lamella is still attached to the sample, figure 13e.

- Lift-out and attach to TEM grid.

Insert the manipulator and Pt GIS carefully to drive the needle as close to the lamella as possible and deposit Pt to weld the needle to the lamella.

First cut the lamella free before lift out of the lamella, figure 13f. The lamella can be lift out and transferred to the TEM grid.

- Lamella thinning.

After attaching the lamella to the TEM grid, figure 13g, the lamella should be thinned. This happens in steps, first thin it down to approximately 800 nm, next 300 nm and at last 80 nm. In the case of the mica lamella the last step is approximately 200 nm, figure 13h.

- Low kV cleaning.

At last low kV cleaning can be done to minimize the ion beams radiation damage, hence to remove surface amorphization. The lamella is obtained, figure 13i.

For more specific preparation details see reference [20].

(22)

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

Figure 13: The process of making a lamella

(23)

7.2 Diffraction pattern in the squares

In figure 14, the diffraction pattern per dose can be seen.

(a) Dose = 3.4 ∗ 10⁻⁶ pC/µm² (b) Dose = 1.7 ∗ 10⁻⁶pC/µm²

(c) Dose = 8.6 ∗ 10⁻⁷pC/µm² (d) Dose = 4.1 ∗ 10⁻⁷pC/µm²

(e) Dose = 2.2 ∗ 10⁻⁷pC/µm² (f) Dose = 0 pC/µm² Figure 14: Diffraction patterns

(24)

7.3 Real space images

In figure 15 the real space TEM images of the boxes in magnitude 500k can be seen. In figure 16 the real space TEM images of the boxes in magnitude 800k

can be seen.

(25)

(26)

7.4 STEM voltage difference

In figure 17 a lamella with boxes can be seen by different voltages ranging from 5 kV to 30 kV electrons. After 15 kV the lamella is clearly illustrated which means that more electrons travel through the lamella than for instance at 5 kV.

(a) 5 kV (b) 10 kV

(c) 15 kV (d) 20 kV

(e) 25 kV (f) 30 kV

Figure 17: Different voltage for imaging the lamella.

(27)

7.5 Matlab script

The following Matlab script, plots a Lorentzian fit through the the central peak intensity data points. Furthermore the central peaks will be aligned to get a clear overview of the increasing of the peaks per increasing dose. Moreover w, A, σw and σA can be obtained for calculating the line integral and the error in the line integral of the central peaks.

1 f u n c t i o n s c r i p t 1 2 3( )

2 % Load p r o f i l e data

3 load(’ P r o f i l e s l i n e s c a n s 20 x1463 . mat ’) ;

4 y1 = ProfileOfFrame2200cmSAEDbox11 + 1 ;

9 y6 = ProfileOfFrame2200cmSAEDbox2background;

10

11 % Default x vector

12 x = 1 : 1 4 6 3 ;

13

14 % Lorentzian f i t o f c e n t r a l peak

15 % b = y0 , xc , w, A; y = y0 + A ∗ (w / (4 ∗ ( x − xc ) ^2 + w

^2) )

16 l o r = @(b, x) b( 1 ) + b( 4) ∗ b(3 ) . / (4 ∗ (x − b( 2 ) ) .^2 + b( 3 ) ^2) ;

17 b0 = [ 0 730 10 1 0 0 0 ] ;

18

19 [b1, ~ , ~ , cov1] = n l i n f i t(x( 7 1 0 : 7 5 0 ) , y1( 7 1 0 : 7 5 0 ) , lor , b0) ;

25

26 %Define e r r o r

27 sigma1 = s q r t(diag(cov1) ) ;

33

34 % More p r e c i s e x vector f o r f i t f u n c t i o n

35 x f i t = 7 1 0 : 0 . 1 : 7 5 0 ;

(28)

36

37 % Plot o f f i t r e s u l t s

38 f i g u r e; hold on;

39 p l o t(x( 7 1 0 : 7 5 0 ) , y1( 7 1 0 : 7 5 0 ) , ’ b∗ ’) ;

40 p l o t(x( 7 1 0 : 7 5 0 ) , y2( 7 1 0 : 7 5 0 ) , ’ r ∗ ’) ;

41 p l o t(x( 7 1 0 : 7 5 0 ) , y3( 7 1 0 : 7 5 0 ) , ’ y∗ ’) ;

42 p l o t(x( 7 1 0 : 7 5 0 ) , y4( 7 1 0 : 7 5 0 ) , ’ g∗ ’) ;

43 p l o t(x( 7 1 0 : 7 5 0 ) , y5( 7 1 0 : 7 5 0 ) , ’ k∗ ’) ;

44 p l o t(x( 7 1 0 : 7 5 0 ) , y6( 7 1 0 : 7 5 0 ) , ’ c∗ ’) ;

45 p l o t(x f i t , l o r(b1, x f i t) ,’b ’) ;

46 p l o t(x f i t , l o r(b2, x f i t) ,’ r ’) ;

47 p l o t(x f i t , l o r(b3, x f i t) ,’ y ’) ;

48 p l o t(x f i t , l o r(b4, x f i t) ,’ g ’) ;

49 p l o t(x f i t , l o r(b5, x f i t) ,’ k ’) ;

50 p l o t(x f i t , l o r(b6, x f i t) ,’ c ’) ;

51 y l a b e l(’ I n t e n s i t y ( counts ) ’) ;

52 x l a b e l(’ P i x e l ’) ;

53 legend(’Box 1 ’,’Box 2 ’,’Box 3 ’,’Box 4 ’,’Box 5 ’,’

Background ’,’ Fit box 1 ’,’ Fit box 2 ’,’ Fit box 3 ’, ’ Fit box 4 ’,’ Fit box 5 ’,’ Fit background ’) ;

54

55 % S h i f t o f d e f a u l t x vector with f i t t e d data

56 x1 = x − b1( 2 ) ;

57 x2 = x − b2( 2 ) ;

58 x3 = x − b3( 2 ) ;

59 x4 = x − b4( 2 ) ;

60 x5 = x − b5( 2 ) ;

61 x6 = x − b6( 2 ) ;

62

63 % Plot o f r e s u l t s

64 f i g u r e; hold on;

65 p l o t(x1, y1) ;

66 p l o t(x2, y2) ;

67 p l o t(x3, y3) ;

68 p l o t(x4, y4) ;

69 p l o t(x5, y5) ;

70 p l o t(x6, y6) ;

71 xlim([ −300 3 0 0 ] )

72 ylim( [ 0 2 3 0 0 ] )

73 y l a b e l(’ I n t e n s i t y ( counts ) ’) ;

74 x l a b e l(’ x−xc ( P i x e l ) ’) ;

75 legend(’Box 1 ’,’Box 2 ’,’Box 3 ’,’Box 4 ’,’Box 5 ’,’ Background ’) ;

76

77 format long

78 79 b1’

80 cov1’

81 sigma1

82 b2’

(29)

83 sigma2

84 b3’

85 sigma3

86 b4’

87 sigma4

88 b5’

89 sigma5

90 b6’

91 sigma6

Investigation of the electron beam sensitivity of mica for electron microscopy characterization

Electron beam sensitivity of mica