*2.2. Analysis Tool Setups*

*Nanomaterials* **2019**, *9*, 1035 is equipped with a bismuth (Bi) liquid metal primary ion gun (LMIG). The primary ion source was used with a short pulse at the voltage of 25 kV (anode) at 45◦. The beam was chopped in 20 ns and focused to about 300 nm in diameter. The ion beam current was ∼13 nA (DC mode). The secondary ions were collected with an extraction lens (biased at −40 V), traveled through the reflection flight tube and identified with a micro-channel plate (MCP) detector. A dose density of primary ions for each measurement was about 6.5 × 10−<sup>13</sup> ions/cm2. The vacuum pressure in the main chamber was kept at about1.0×10−<sup>8</sup> mbar.The SIMS analysis data were acquired using a TOF-SIMS 300R (IONTOF GmbH, Muࡇnster, Germany) in order to analyze both overall film growth and doping levels inside the structures. The tool is equipped with a bismuth (Bi) liquid metal primary ion gun (LMIG). The primary ion source was used with a short pulse at the voltage of 25 kV (anode) at 45°. The beam was chopped in 20 ns and focused to about 300 nm in diameter. The ion beam current was 13 nA (DC mode). The secondary ions were collected with an extraction lens (biased at −40 V), traveled through the reflection flight tube and identified with a micro-channel plate (MCP) detector. A dose density of primary ions for each measurement was about 6.5 × 10−13 ions/cm2. The vacuum pressure in the main chamber was kept at about 1.0 × 10−8 mbar.

 A mass spectrum for each pixel was collected by raster-scanning of the primary ion beam to obtain a mass spectrometry data image across the sample area of interest (150 × 150 μm<sup>2</sup> using TSVs and 102 × 102 μm<sup>2</sup> using LHAR test chips). An electron impact (EI) gas ion source of oxygen (O2 +) was used as a second ion beam source to remove material for profiling electropositive elements in positive SIMS. The energy of the sputter gun was set to 1 kV to concede the image profile of interesting fragments in depth. The average beam current was 0.23 μA. The initial velocity of secondary ions was provided by an accelerator into the drift tube with a potential of 2 kV. This velocity distribution was implemented for the mass separation by flight time analysis. A mass resolution of 8300 (m/Δm) at 28 amu can be accomplished with the flow time transformed into atomic mass units (amu). For the image analysis, data were acquired from 400 × 400 μm<sup>2</sup> by an ion beam raster of 256 × 256 pixels in asaw-toothmode.ImageanalysisisconductedinSurfaceLabv6.3(developedatIONTOFGmbH).A mass spectrum for each pixel was collected by raster-scanning of the primary ion beam to obtain a mass spectrometry data image across the sample area of interest (150 × 150 μm2 using TSVs and 102 × 102 μm2 using LHAR test chips). An electron impact (EI) gas ion source of oxygen (O2+) was used as a second ion beam source to remove material for profiling electropositive elements in positive SIMS. The energy of the sputter gun was set to 1 kV to concede the image profile of interesting fragments in depth. The average beam current was 0.23 μA. The initial velocity of secondary ions was provided by an accelerator into the drift tube with a potential of 2 kV. This velocity distribution was implemented for the mass separation by flight time analysis. A mass resolution of 8300 (m/ Δm) at 28 amu can be accomplished with the flow time transformed into atomic mass units (amu). For the image analysis, data were acquired from 400 × 400 μm2 by an ion beam raster of 256 × 256 pixels in a saw-tooth mode. Image analysis is conducted in SurfaceLab v6.3 (developed at IONTOF GmbH).

 For step coverage analysis we used SEM by Hitachi S4800 (Hitachi, Japan) with an electron beam source energy at 10 kV using a back-scattered detector. For higher lateral resolution on the TSV cross-section, TEM observation was carried out for the structural analysis in localized areas at the bottom of the TSV sidewalls using FEI Tecnai F20 (Hillsboro, OR, USA). The acceleration energy of the electron beam was 200 keV. The process of lamella preparation for TEM observation is explained in Appendix A. For step coverage analysis we used SEM by Hitachi S4800 (Hitachi, Japan) with an electron beam source energy at 10 kV using a back-scattered detector. For higher lateral resolution on the TSV crosssection, TEM observation was carried out for the structural analysis in localized areas at the bottom of the TSV sidewalls using FEI Tecnai F20 (Hillsboro, OR, USA). The acceleration energy of the electron beam was 200 keV. The process of lamella preparation for TEM observation is explained in Appendix A.

## **3.Results andDiscussion3. Results and Discussion**

#### *3.1.VerticalHigh AspectRatioStructures 3.1. Vertical High Aspect Ratio Structures*

As mentioned earlier, one of the problems with characterizing the elemental distribution in TSVs is the geometry of the structures in comparison with planar samples. The surface topography can cause undesirable artifacts in the SIMS image and spectra profile which significantly inhibits interpretation and quantification of data. As a consequence, surface evaluation and depth analysis with microscale topography are a considerable challenge due to analytical instrumental limitation. To study the inside of TSVs, the samples were cut and mounted on the sample holder in such a way that the cross sectioned TSV sidewalls face the analyzer. Figure 3 demonstrates the analysis area and sputter zone in the cross-sectionalview. As mentioned earlier, one of the problems with characterizing the elemental distribution in TSVs is the geometry of the structures in comparison with planar samples. The surface topography can cause undesirable artifacts in the SIMS image and spectra profile which significantly inhibits interpretation and quantification of data. As a consequence, surface evaluation and depth analysis with microscale topography are a considerable challenge due to analytical instrumental limitation. To study the inside of TSVs, the samples were cut and mounted on the sample holder in such a way that the cross sectioned TSV sidewalls face the analyzer. Figure 3 demonstrates the analysis area and sputter zone in the cross-sectional view. 

**Figure 3.** Cross-sectional images showing (**a**) an SEM micrograph of TSVs and (**b**) a camera snapshot of a mounted sample (without Cu plating) showing the side wall with the crater size and analysis area, as highlighted in the image, during the ToF-SIMS measurement.

A 3D reconstruction of the ToF-SIMS data was used to map the elemental distribution. The resulting image intensity is influenced by the effect of sample surface orientation or surface topography of the analysis area on the trajectory of the primary ion beam. Sputter yields not only depend on the primary ion energy but also the angle of incidence [2,25,26]. The angle of incidence changes depending on of the surface topography [3]. This induces changes in the energy distribution of the emitted secondary ions, the sputtering yield, and the angular distribution of ejected fragments (based on the emission angle from the surface normal) [27,28]. This artifact can cause falsified variations in image intensities (see Figure 4). A 3D reconstruction of the ToF-SIMS data was used to map the elemental distribution. The resulting image intensity is influenced by the effect of sample surface orientation or surface topography of the analysis area on the trajectory of the primary ion beam. Sputter yields not only depend on the primary ion energy but also the angle of incidence [2,25,26]. The angle of incidence changes depending on of the surface topography [3]. This induces changes in the energy distribution of the emitted secondary ions, the sputtering yield, and the angular distribution of ejected fragments (based on the emission angle from the surface normal) [27,28]. This artifact can cause falsified variations in image intensities (see Figure 4). 

**Figure 4.** Schematic representation of a TSVs cross-section. Primary ion beams strike the sample's top surface at an incident angle of 45° (red arrows). The *n*1, *n*2, and *n*3 denote the axes normal to each surface (blue colored planes). Each of these surfaces depict the relative angle between the incident ion beams and the inner surface of the TSV. **Figure 4.** Schematic representation of a TSVs cross-section. Primary ion beams strike the sample's top surface at an incident angle of 45◦ (red arrows). The *n*1, *n*2, and *n*3 denote the axes normal to each surface (blue colored planes). Each of these surfaces depict the relative angle between the incident ion beams and the inner surface of the TSV.

The acceptance angle of the primary ions (red arrows) with respect to the sample normal of the left inner surface (*n*3), depicted in Figure 4, is limited in comparison to the middle and right inner surface (*n*2, *n*1). Consequently, the detection of secondary ions resulting from the right inner surface have a higher probability to reach the mass analyzer. Whereas, the secondary ions from the left inner surface can hardly reach the analyzer. The image profile of a 25 nm thick film of Co in the TSV sputtered using a 1 kV oxygen-ion beam, is shown in Figure 5a. Cobalt is visible as 59Co+ (*m*/*z* = 58.93) throughout the TSV. Figure 6a, shows that Co is completely corroded after electroplating. As illustrated in Figures 5a and 6a, the same constructed layout, as in Figure 4, is built from ToF-SIMS analysis. The samples' orientations, with respect to the primary ion beams, and the analyzer perfectly agree with the schematic representation in Figure 4. The acceptance angle of the primary ions (red arrows) with respect to the sample normal of the left inner surface (*n*3), depicted in Figure 4, is limited in comparison to the middle and right inner surface (*<sup>n</sup>*2, *n*1). Consequently, the detection of secondary ions resulting from the right inner surface have a higher probability to reach the mass analyzer. Whereas, the secondary ions from the left inner surface can hardly reach the analyzer. The image profile of a 25 nm thick film of Co in the TSV sputtered using a 1 kV oxygen-ion beam, is shown in Figure 5a. Cobalt is visible as 59Co+ (*m*/*z* = 58.93) throughout the TSV. Figure 6a, shows that Co is completely corroded after electroplating. As illustrated in Figures 5a and 6a, the same constructed layout, as in Figure 4, is built from ToF-SIMS analysis. The samples' orientations, with respect to the primary ion beams, and the analyzer perfectly agree with the schematic representation in Figure 4.

**Figure 5.** Images shown here are of the TSV before Cu electroplating. In (**a**) the 3D ToF-SIMS map of Si and Co signal from the bottom is shown and in (**b**) a TEM image from the bottom sidewall of the TSV showing the TaN barrier the cobalt film and the preparation fill.

**Figure 6.** Images shown here are of the TSV after Cu electroplating. In (**a**) the 3D ToF-SIMS map of Si, Co and Cu signals from the bottom and in (**b**) a TEM image from the bottom sidewall of the TSV is shown.

In Figures 5a and 6a, the geometrical influence on 3D imaging is pronounced for the trench sidewalls. A higher contrast is observed for Co and Cu of the sidewalls which face the primary ion gun. This confirms the influence of surface topography on SIMS 3D image reconstruction. In most SIMS instruments, the incident angle of the primary ions (viewing direction) creates an angle between 30◦ and 60◦. As a result, the shadowing effects of the surface geometry are unavoidable [3]. Although, different approaches are suggested by Lee et al. such as sample alignment, changing extraction voltage, using cluster ion beam, and adjusting extraction delay to reduce the shadow effect [29]. However, this effect could not be eliminated entirely.

In addition, we compare the ToF-SIMS results with TEM micrographs which support the findings of ToF-SIMS shown in Figures 5b and 6b. Both TEM lamellas are lifted out from the bottom of the TSVs. The Pt and C that are marked in the figures are used as the protection layers in the process of lamella preparation for TEM to protect the top surface from unwanted surface damage by FIB milling and amorphization with high energetic ions. Comparing the results from TEM with ToF-SIMS, 3D images determined the Co corrosion at the bottom of TSV. The TEM image in Figure 5b shows the 25 nm thick Co seed layer on top of the TaN, which matches with ToF-SIMS analysis results. After Cu electroplating, it is not possible to detect Co species at the bottom of the TSV sidewall. Besides, Cu signal intensity is very low, which supports the idea of complete corrosion at the bottom of the TSV. To check the precision of this observation, a new lamella was prepared from the sample after electroplating. The results in Figure 6b support the assumption of Co thin film corrosion and failure on Cu deposition.

#### *3.2. Lateral High Aspect Ratio Structures*

One type of ferroelectric based non-volatile memory is the 1T-1C ferroelectric random-access memory (FRAM) where the memory state is stored in a ferroelectric capacitor. The readout current through the access transistor becomes proportional to the ferroelectric remnant polarization. In order to improve the sense current, a common 3D-like structure is utilized where the area factor enhances the total polarization charge [30]. Various dopants influence the stabilization of the ferroelectric phase in hafnia thin films, for example, silicon (Si) [17,31,32], and such influence becomes more decisive in deep trench structures while the dopants concentration may change inside the HAR structure. Remarkably, it is more critical in low-doped materials, where such a low Si content is necessary, while a small deviation results in strong change in the stabilized phase [33]. As compared to planar film deposition, effects on the layer composition and crystal structure are expected to be different for HAR structures. Therefore, a technique to quantify the Si concentration in ferroelectric films deposited in 3D structures is crucial for optimizing the thin film deposition process of FRAM applications.

 structures. Therefore, a technique to quantify the Si concentration in ferroelectric films deposited in 3D structures is crucial for optimizing the thin film deposition process of FRAM applications.

Due to the small dimensions of 3D structures it becomes crucial to reach an attainable lateral resolution of different analytical techniques. In ToF-SIMS 3D imaging, the dimension of collision cascade and surface damage define the limitation that one could reach in the lateral resolution. Consequently, the useful lateral resolution of SIMS imaging relies on the primary ion beam focus and ionization efficiency of the surface analysis [34]. In order to reach the maximal lateral resolution, it is a compromise between acquisition time (number of ions per shot), mass resolution (pulse length) and lateral resolution (spot size) is unavoidable. To succeed in high mass resolution, short pulses in the order of < 1 ns are required. Producing short pulses with higher energy cost deterioration of the focus in the range of 3–10 μm. In contrast, to achieve a lateral resolution in the range of 200–300 nm, it is required to increase the pulse length larger than 150 ns whilst decreasing the mass resolution. In spite of such a trade-off, it is also possible to simultaneously obtain high spatial and high mass resolution as described by Vanbellingen et al. by applying delay extraction of secondary ions [35]. Due to the small dimensions of 3D structures it becomes crucial to reach an attainable lateral resolution of different analytical techniques. In ToF-SIMS 3D imaging, the dimension of collision cascade and surface damage define the limitation that one could reach in the lateral resolution. Consequently, the useful lateral resolution of SIMS imaging relies on the primary ion beam focus and ionization efficiency of the surface analysis [34]. In order to reach the maximal lateral resolution, it is a compromise between acquisition time (number of ions per shot), mass resolution (pulse length) and lateral resolution (spot size) is unavoidable. To succeed in high mass resolution, short pulses in the order of < 1 ns are required. Producing short pulses with higher energy cost deterioration of the focus in the range of 3–10 μm. In contrast, to achieve a lateral resolution in the range of 200–300 nm, it is required to increase the pulse length larger than 150 ns whilst decreasing the mass resolution. In spite of such a trade-off, it is also possible to simultaneously obtain high spatial and high mass resolution as described by Vanbellingen et al. by applying delay extraction of secondary ions [35].

Figure 7a,b show SEM micrographs of VHAR structures with 20 nm of HSO deposited on them. The non-planar surface geometry of the collected samples' cross sections influences the secondary ion yield and restricts the number of ions that can be created out of the sputtered area. In addition, the surface topography has an effect on the ultimate mass resolution, which arises from the slight differences in the extraction field [36]. In low-intensity imaging, many small features are affected by artifacts of counting statistics. This limits the total amount of achievable information and results in only a few number of counts of interest area per pixel which makes it crucial to properly evaluate the results from planer surfaces. Figure 7a,b show SEM micrographs of VHAR structures with 20 nm of HSO deposited on them. The non-planar surface geometry of the collected samples' cross sections influences the secondary ion yield and restricts the number of ions that can be created out of the sputtered area. In addition, the surface topography has an effect on the ultimate mass resolution, which arises from the slight differences in the extraction field [36]. In low-intensity imaging, many small features are affected by artifacts of counting statistics. This limits the total amount of achievable information and results in only a few number of counts of interest area per pixel which makes it crucial to properly evaluate the results from planer surfaces.

**Figure 7.** SEM cross-section of 20 nm Si-doped HfO2 (HSO) deposited on vertical high aspect ratio (VHAR) structures are shown in (**<sup>a</sup>**–**d**). In (**e**), a side-view SEM micrograph of PillarHall lateral high aspect ratio (LHAR) test structure (membrane width 20 μm) is shown and in (**f**), a top-view SEM micrograph of Si-doped HfO2 grown at 280 °C in LHAR structure (PillarHall) is shown. The membrane has been peeled off (membrane width > 500 μm). The central cavity opening for precursors flow shows with "W" (see Figure 2a). The diffusion depth of the ALD process can be examined directly by ToF-SIMS. **Figure 7.** SEM cross-section of 20 nm Si-doped HfO2 (HSO) deposited on vertical high aspect ratio (VHAR) structures are shown in (**<sup>a</sup>**–**d**). In (**e**), a side-view SEM micrograph of PillarHall lateral high aspect ratio (LHAR) test structure (membrane width 20 μm) is shown and in (**f**), a top-view SEM micrograph of Si-doped HfO2 grown at 280 ◦C in LHAR structure (PillarHall) is shown. The membrane has been peeled off (membrane width > 500 μm). The central cavity opening for precursors flow shows with "W" (see Figure 2a). The diffusion depth of the ALD process can be examined directly by ToF-SIMS.

To improve image analysis, we facilitate the characterization procedure of HAR structures on the LHAR to create a suitable technique on imaging elemental distribution in non-planar structures with ToF-SIMS. Figure 7e,f illustrate an LHAR structure via SEM micrographs. The chips (PillarHall) can easily be brought on to a carrier wafer in the ALD chamber. After the deposition process, the membrane is peeled off using an adhesive tape. To improve image analysis, we facilitate the characterization procedure of HAR structures on the LHAR to create a suitable technique on imaging elemental distribution in non-planar structures with ToF-SIMS. Figure 7e,f illustrate an LHAR structure via SEM micrographs. The chips (PillarHall) can easily be brought on to a carrier wafer in the ALD chamber. After the deposition process, the membrane is peeled off using an adhesive tape.

*Nanomaterials* **2019**, *9*, 1035 *Nanomaterials* **2019**, *9*, x FOR PEER REVIEW 9 of 14

**Figure 8.** Images showing (**a**) snapshot using a microscope from the PillarHall after membrane 

**Figure 8.** Images showing (**a**) snapshot using a microscope from the PillarHall after membrane removal (top view), (**b**) a superimposed image with ToF-SIMS of deposited region, and (**<sup>c</sup>**–**<sup>e</sup>**) an RGB representation of 16O+ (*m*/*<sup>z</sup>* = 15.994), 28Si+ (*m*/*<sup>z</sup>* = 27.978), and 180Hf + (*m*/*<sup>z</sup>* = 179.949). **Figure 8.** Images showing (**a**) snapshot using a microscope from the PillarHall after membrane removal (top view), (**b**) a superimposed image with ToF-SIMS of deposited region, and (**<sup>c</sup>**–**<sup>e</sup>**) an RGB representation of 16O+ (*m*/*z* = 15.994), 28Si+ (*m*/*z* = 27.978), and 180Hf + (*m*/*z* = 179.949). removal (top view), (**b**) a superimposed image with ToF-SIMS of deposited region, and (**<sup>c</sup>**–**<sup>e</sup>**) an RGB representation of 16O+ (*m*/*<sup>z</sup>* = 15.994), 28Si+ (*m*/*<sup>z</sup>* = 27.978), and 180Hf + (*m*/*<sup>z</sup>* = 179.949). Figure 8b represents the total ion image spectrum of the acquired area associated with the 16O+,

Figure 8b represents the total ion image spectrum of the acquired area associated with the 16O+, 28Si+, and 180Hf+ signals. In Figures 8c–e, each elemental signal is color coded in red, green and blue for 16O+, 28Si+, and 180Hf+, respectively. The green dots in Figures 8b and 8d are the holes left by the pillars after membrane removal. Figure 9a illustrates the volumetric view by 3D data analysis of the distribution of 16O+ (*m*/*<sup>z</sup>* = 15.994), 28Si+ (*m*/*<sup>z</sup>* = 27.978) and 180Hf+ (*m*/*<sup>z</sup>* = 179.949) over a 102 × 102 μm<sup>2</sup> area. To assess the deposited material in the LHAR structure with respect to the depth, the region of interests (ROIs) with a dimension of 2 × 30 μm2 were defined as depicted in Figure 9b. For each ROI, the 28Si+ and 180Hf+ species total counts from emitted secondary ions are integrated, where the influence of the silicon substrate is excluded based on the 28Si+ depth profile variation from 16% to 84% on minimum 28Si+/180Hf+ ratio (Figure 10\_ROI.01). We are assuming a homogeneous spatial distribution of Si+ and Hf+ on each ROI (2 × 30 μm2). Figure 8b represents the total ion image spectrum of the acquired area associated with the 16O<sup>+</sup>, 28Si<sup>+</sup>, and 180Hf+ signals. In Figure 8c–e, each elemental signal is color coded in red, green and blue for 16O<sup>+</sup>, 28Si<sup>+</sup>, and 180Hf<sup>+</sup>, respectively. The green dots in Figure 8b,d are the holes left by the pillars after membrane removal. Figure 9a illustrates the volumetric view by 3D data analysis of the distribution of 16O+ (*m*/*z* = 15.994), 28Si+ (*m*/*z* = 27.978) and 180Hf+ (*m*/*z* = 179.949) over a 102 × 102 μm<sup>2</sup> area. To assess the deposited material in the LHAR structure with respect to the depth, the region of interests (ROIs) with a dimension of 2 × 30 μm<sup>2</sup> were defined as depicted in Figure 9b. For each ROI, the 28Si+ and 180Hf+ species total counts from emitted secondary ions are integrated, where the influence of the silicon substrate is excluded based on the 28Si+ depth profile variation from 16% to 84% on minimum 28Si+/180Hf+ ratio (Figure 10\_ROI.01). We are assuming a homogeneous spatial distribution of Si+ and Hf+ on each ROI (2 × 30 μm2). 28Si+, and 180Hf+ signals. In Figures 8c–e, each elemental signal is color coded in red, green and blue for 16O+, 28Si+, and 180Hf+, respectively. The green dots in Figures 8b and 8d are the holes left by the pillars after membrane removal. Figure 9a illustrates the volumetric view by 3D data analysis of the distribution of 16O+ (*m*/*<sup>z</sup>* = 15.994), 28Si+ (*m*/*<sup>z</sup>* = 27.978) and 180Hf+ (*m*/*<sup>z</sup>* = 179.949) over a 102 × 102 μm<sup>2</sup> area. To assess the deposited material in the LHAR structure with respect to the depth, the region of interests (ROIs) with a dimension of 2 × 30 μm2 were defined as depicted in Figure 9b. For each ROI, the 28Si+ and 180Hf+ species total counts from emitted secondary ions are integrated, where the influence of the silicon substrate is excluded based on the 28Si+ depth profile variation from 16% to 84% on minimum 28Si+/180Hf+ ratio (Figure 10\_ROI.01). We are assuming a homogeneous spatial distribution of Si+ and Hf+ on each ROI (2 × 30 μm2). 

(**a**) (**b**) **Figure 9.** (**a**) A volumetric view of the distribution of 16O+ (*m*/*<sup>z</sup>* = 15.994), 28Si+ (*m*/*<sup>z</sup>* = 27.978) and 180Hf+ (*m*/*<sup>z</sup>* = 179.949) over a 102 × 102 μm2 field of view from the deposited region after peeling off the membrane. (**b**) A defined ROI with dimension of 2 × 30 μm2 to integrate total counts of the interested **Figure 9.** (**a**) A volumetric view of the distribution of 16O+ (*m*/*<sup>z</sup>* = 15.994), 28Si+ (*m*/*<sup>z</sup>* = 27.978) and 180Hf+ (*m*/*<sup>z</sup>* = 179.949) over a 102 × 102 μm2 field of view from the deposited region after peeling off the membrane. (**b**) A defined ROI with dimension of 2 × 30 μm2 to integrate total counts of the interested area from Si and Hf. **Figure 9.** (**a**) A volumetric view of the distribution of 16O+ (*m*/*z* = 15.994), 28Si+ (*m*/*z* = 27.978) and 180Hf+ (*m*/*z* = 179.949) over a 102 × 102 μm<sup>2</sup> field of view from the deposited region after peeling off the membrane. (**b**) A defined ROI with dimension of 2 × 30 μm<sup>2</sup> to integrate total counts of the interested area from Si and Hf.

area from Si and Hf. The ToF-SIMS analysis results of the HSO material in the LHAR structure are shown in Figure 10. The integrated counts for the 180Hf+ and 28Si+ species are plotted with respect to the depth inside the test structure. Due to the geometry of the sample, the analysis starts at a minimal distance of 5 μm from the inlet slit. The absolute intensity is not constant as the sputter rate exhibits local variations (not shown here). However, the ratio of the 180Hf+ and 28Si+ signals is consistent over a depth range The ToF-SIMS analysis results of the HSO material in the LHAR structure are shown in Figure 10. The integrated counts for the 180Hf+ and 28Si+ species are plotted with respect to the depth inside the test structure. Due to the geometry of the sample, the analysis starts at a minimal distance of 5 μm from the inlet slit. The absolute intensity is not constant as the sputter rate exhibits local variations (not shown here). However, the ratio of the 180Hf+ and 28Si+ signals is consistent over a depth range from 5 μm to 30 μm. The ToF-SIMS analysis results of the HSO material in the LHAR structure are shown in Figure 10. The integrated counts for the 180Hf+ and 28Si+ species are plotted with respect to the depth inside the test structure. Due to the geometry of the sample, the analysis starts at a minimal distance of 5 μm from the inlet slit. The absolute intensity is not constant as the sputter rate exhibits local variations (not shown here). However, the ratio of the 180Hf+ and 28Si+ signals is consistent over a depth range from 5 μm to 30 μm.

from 5 μm to 30 μm. 

**.** 

**Figure 10.** The 28Si+, 180Hf+ and the ratio signals of the HfO2 layer obtained by ToF-SIMS analysis of the LHAR test structure with respect to the diffusion depth. **Figure 10.** The 28Si<sup>+</sup>, 180Hf+ and the ratio signals of the HfO2 layer obtained by ToF-SIMS analysis of the LHAR test structure with respect to the di ffusion depth.

Total area counts of the 28Si+ and 180Hf+ signals were calculated from the defined areas (ROIs) of the 3D reconstruction (as seen in Figure 9b). A constant 28Si+ to 180Hf+ signal ratio is observed over the depth range until *c.a.* 30 μm (deviation of 5%) which corresponds to an aspect ratio up to 1:50 with uniform dopant concentration. At a depth higher than 35 μm, the Si content rises. This can be due to the Si substrate influence. However, up to *c.a.* 30 μm, the step coverage (or depth penetration) of 3DMAS/O3 (SiO2) is as good as TEMAHf/O3 (HfO2) at the chosen process conditions. Total area counts of the 28Si+ and 180Hf+ signals were calculated from the defined areas (ROIs) of the 3D reconstruction (as seen in Figure 9b). A constant 28Si+ to 180Hf+ signal ratio is observed over the depth range until *c.a.* 30 μm (deviation of 5%) which corresponds to an aspect ratio up to 1:50 with uniform dopant concentration. At a depth higher than 35 μm, the Si content rises. This can be due to the Si substrate influence. However, up to *c.a.* 30 μm, the step coverage (or depth penetration) of 3DMAS/O3 (SiO2) is as good as TEMAHf/O3 (HfO2) at the chosen process conditions. **4. Conclusions**

**4. Conclusions**  This study aimed to illustrate a novel method to address the challenges of elemental analysis in high aspect ratio structures and to provide a practical guideline on the effect of surface topography on ToF-SIMS. As shown, the main difficulty in the interpretation of non-planar structures is to reach optimal material removal and sensitivity for a given material. As shown in the first study, on the vertical TSVs, the image intensity is affected by the surface topography and results in an inaccurate representation of 3D imaging. Since the sputter yield is a function of the primary ion beam incident angle (incident at θ = 45° to the sample normal for a flat sample), the geometrical shadowing effects are inevitable. In addition, we used lateral high aspect ratio (LHAR) test structures (PillarHall™) as a new platform to perform precise measurements of hafnia doped thin films. In contrast to the common vertical HAR structures, the combination of ToF-SIMS 3D imaging and LHAR test chips leads to a constructed footprint of the composition of the deposited thin film without being influenced by the sputter yield and the surface topography. By integrating the total data points of the ROIs, semi-quantitative calculations on the ratio of elemental distribution over the depth of the LHAR structures were calculated. In conclusion, surface topographic effects could be eliminated by using LHAR structures, allowing the separation of chemical variations within the analysis area. Further investigations are planned to precisely compare the stoichiometry of the deposited material in the planar structures with the deposited film in the depth of the LHAR structures by coupling results This study aimed to illustrate a novel method to address the challenges of elemental analysis in high aspect ratio structures and to provide a practical guideline on the e ffect of surface topography on ToF-SIMS. As shown, the main di fficulty in the interpretation of non-planar structures is to reach optimal material removal and sensitivity for a given material. As shown in the first study, on the vertical TSVs, the image intensity is a ffected by the surface topography and results in an inaccurate representation of 3D imaging. Since the sputter yield is a function of the primary ion beam incident angle (incident at θ = 45◦ to the sample normal for a flat sample), the geometrical shadowing e ffects are inevitable. In addition, we used lateral high aspect ratio (LHAR) test structures (PillarHall ™) as a new platform to perform precise measurements of hafnia doped thin films. In contrast to the common vertical HAR structures, the combination of ToF-SIMS 3D imaging and LHAR test chips leads to a constructed footprint of the composition of the deposited thin film without being influenced by the sputter yield and the surface topography. By integrating the total data points of the ROIs, semi-quantitative calculations on the ratio of elemental distribution over the depth of the LHAR structures were calculated. In conclusion, surface topographic e ffects could be eliminated by using LHAR structures, allowing the separation of chemical variations within the analysis area. Further investigations are planned to precisely compare the stoichiometry of the deposited material in the planar structures with the deposited film in the depth of the LHAR structures by coupling results from ToF-SIMS and X-ray photoelectron spectroscopy (XPS).

from ToF-SIMS and X-ray photoelectron spectroscopy (XPS). **Author Contributions:** Investigation and Methodology conducted by A.M.K., C.M. and N.H. Writing the original draft by A.M.K. All authors contributed to the review and editing. LHAR test structures from M.U. and R.L.P. TSVs are provided by S.E. Thin films depositions conducted by S.E. and C.M. ToF-SIMS, SEM and TEM conducted by A.M.K. Results visualization done by A.M.K. The supervision of the project was by N.H. and W.W. **Author Contributions:** Investigation and Methodology conducted by A.M.K., C.M. and N.H. Writing the original draft by A.M.K. All authors contributed to the review and editing. LHAR test structures from M.U. and R.L.P. TSVs are provided by S.E. Thin films depositions conducted by S.E. and C.M. ToF-SIMS, SEM and TEM conducted by A.M.K. Results visualization done by A.M.K. The supervision of the project was by N.H. and W.W. **Funding:** This research was funded by the German federal ministry of education and research (BMBF) within the project name "VEProSi", No. 13XP5020C and European Regional Development Fund (EFRE) and the Free State of Saxony under the project name "CONSIVA", No. 100273858.

**Funding:** This research was funded by the German federal ministry of education and research (BMBF) within the project name ''VEProSi", No. 13XP5020C and European Regional Development Fund (EFRE) and the Free State of Saxony under the project name "CONSIVA", No. 100273858. **Acknowledgments:** The author gratefully thanks Jennifer Emara for reviewing and commenting on the manuscript. **Conflicts of Interest:** The authors declare no conflict of interest.

**Acknowledgements:** The author gratefully thanks Jennifer Emara for reviewing and commenting on the manuscript. 

*Nanomaterials* **2019**, *9*, 1035 **Conflicts of Interest:** The authors declare no conflict of interest. **Conflicts of Interest:** The authors declare no conflict of interest.
