“Report on TEM and FIB based microscopy studies …...phase shift, was estimated from the phase...

Project full title: " Nanowires for solid state lighting " Project acronym: NWs4LIGHT

Grant agreement no: 280773

Deliverable D2.2:

• “Report on TEM and FIB based microscopy studies of NW-LEDs”

Responsible authors: Jakob Birkedal-Wagner, DTU and Anders Gustafsson, ULUND

Summary

Transmission electron microscopy (TEM) and focussed ion beam (FIB) microscopy of III-V nanowires for NW LEDs has been performed at DTU and ULUND. The main findings are:

• Presence of unexpected electrostatic potential steps across the core/under-layer of nanowires

• Qualitative measurements of the electrostatic potential steps by means of electron holography are not corresponding to intended dopant profile.

• The measured built-in potential across the p-shell/n-core was smaller than the theoretical value

• Quantitative measurements of the built-in potential over the pn junction is giving 0.3V. According to the dopant concentration it should be 3V

• In situ measurements of built-in potential vs. temperature shows linear behaviour.

• Use of electron holography of electrically biased single nanowires in-situ in TEM for measuring the electrical resistivity of nanowires

• We gave demonstrated that electron holography is possible on biased NWs. Here the resistivity of a GaN NW is measured a proof-of-concept.

• Connection between misfit dislocations and stacking faults revealed by focused ion beam milling

• FIB has been used on InN QDs on GaN NWs to optimize the geometry for TEM.

• InN dots on GaN NWs are studied showing stacking faults on misfit dislocations at the interface.



D2.2 TEM and FIB based studies of NW-LEDs Presence of unexpected electrostatic potential steps across the core/under-layer of nanowiresElectron holography is used here to measure the electrostatic potential of GaN NWs on the local scale.

Electron holography enables us to measure the shift induced in the phase of the electron beam as passed through a transmission electron microscope (TEM) specimen. This phase shift, 𝜑(𝑥, 𝑦), is proportional to the distribution of electrostatic potentials, 𝑉(𝑥,𝑦, 𝑧), within the specimen. In the absence of dynamical diffraction and magnetic field, and if it is assumed that the electrostatic potential variation along the beam direction, 𝑧, is negligible, the relation between the phase shift and the electrostatic potential can be simplified to: 𝜑(𝑥,𝑦) = 𝐶𝐸 𝑉(𝑥,𝑦). 𝑡(𝑥,𝑦) (1) where 𝐶𝐸 is a constant depending on the accelerating voltage of the microscope and 𝑡(𝑥,𝑦) represents the specimen thickness.

Electron holography of doped GaN wire heterostructures was used to measure the electrostatic potentiel. Figure 1 shows typical reconstructed off-axis electron holography phase images from cross sectional GaN wires. The sample was cut and prepared using focused ion beam (FIB) in order to optimize the sample geometry for the experiments.

In these phase images, brighter regions represent areas with higher electrostatic potentials and darker regions correspond to areas with lower potentials. Red arrows are pointing to regions with unexpected potential drops. It can be seen that there is a potential drop between the core and under-layer and another unexpected potential drop at the very centre of the core.

It was expected to have a slightly higher potential at the core than the under-layer due to the difference in the dopant concentrations in these regions, but the potential on the core was lower than the under-layer. Also two additional potential drops, between the core and under-layer and at the very centre of the core, were observed.



Figure 1. (a) Structure of investigated nanowires consisting of n-type (Si doped) cores and under-layers with the nominal dopant concentrations of 1×1019 cm-3 and 5×1018 cm-3 respectively. (b) and (c) Typical reconstructed off-axis electron holographic phase images from different cross-sectional views prepared using focused ion beam. (d) Schematic illustration of what we expected and what we observed. Electron holography is quantitative. Figure 2 shows the results of electron holography of GaN wires used for quantitative measurement of the built-in potential across the pn-junction. The expected built-in potential across the shell/under-layer pn junction is about 3V based on the intentinal doping. In Fig 2.1b the dopant contrast, brighter on the p-side shell and darker on the n-side under-layer and core, is used for finding the optimum FIB based specimen preparation conditions. To minimize the FIB damage and maximize the dopant contrast, the specimen surface is polished with low voltage (2 KeV) FIB. In Fig 2.1c the reconstructed electron holography phase image clearly shows the pn junction between the shell and under-layer. The p-type shell is darker than the n-type under-layer as it has a lower electrostatic potential. The quantitative results of the built-in potential measurement is showed in Fig. 2.1d. The measured built-in potential across the shell/under-layer is only 0.3 V, significantly smaller than expected 3V. Possible explanations for this large disagreement are: a) The dopant atoms have not been activated. b) The built-in potential is changed during the TEM specimen preparation. c) The built-in potential is changed because of the electron irradiation during electron holography. In order to examine whether the FIB preparation is responsible for the small measured built-in potential, electron holography was used on a full wire. Figure 2.2 shows the results. The measured built-in potential is 0.4V, only slightly larger than the measured built-in potential from the FIB-prepared specimen. This shows that the smallness of measured built-in potential



in FIB-prepared cross-sectional specimen cannot be solely due to FIB-damage and specimen preparation. In situ annealing of the NWs was performed in order to investigate the temperature dependence of the built-in potential. The temperature depence on the measured built-in potential is shown in Fig 2.3. Before annealing at room temperature the measured built-in potential is 0.3 V. By elevating the temperature and annealing the specimen for one hour at each temperature, for activating possible inactive dopants, the built-in potential increased linearly with the temperature. However, after cooling down to the room temperature the measured potential was only slightly larger than 0.3V which was measured before annealing. This experiment shows that inactivation of dopants also cannot be the reason for the smallness of the measured built-in potential. Therefore, it is very likely that secondary electron emission and electron-hole pair generation as a result of electron irradiation are being the reasons for the smallness of measured built-in potential using electron holography. This needs further investigation.

Figure 2. (1a) Nanowire structure: the structure of core-shell GaN nanowires, consisting of p-type (Mg-doped with Mg to Ga ratio of 3.6%) shell and unintentionally n-type (u-type) core and under-layer, examined using off-axis electron holography. (1b) FIB prepared TEM specimen: secondary electron image of FIB-prepared cross-sectional TEM specimen. (1c) Phase image: reconstructed holographic phase image from a FIB-prepared cross-



sectional nanowire. (1d) Electrostatic potential profile: the electrostatic potential profile extracted from the box shown in (1c). (2a) TEM specimen from intact single nanowires: a single intact nanowire detached and lifted up from the substrate using a micromanipulator and placed on a Cu grid (inset) using only electrostatic forces. (2b) Phase image: reconstructed phase image from an intact nanowire. (2c) Phase image after removing the thickness effect: the pn junction across the shell/under-layer can be seen in this image. (2d) Potential profile: the potential profile extracted from the box in (2c) across the pn junction. (3) In-situ annealing: reconstructed phase images from a FIB-prepared nanowire (3a) before annealing at room temperature, (3b) at 600oC and (3c) after annealing at room temperature. (3d) Measured built-in potential as a function of temperature. A nanowire with the length of L, cross-sectional area of A and electrical resistivity of 𝜌 would have the electrical resistance of R = ρL/A. If the current 𝐼 is passed through this nanowire, the potential drop along ΔL length of the nanowire would be:

𝛥𝑉 = 𝜌∆𝐿𝐴𝐼

(2)

Substituting equation (2) into (1) results in:

𝛥𝜑 = 𝐶𝐸 𝜌𝛥𝐿𝐴𝐼. 𝑡

(3)

Therefore, the phase shift measured along a nanowire using electron holography increases linearly with the current, and the resistivity of nanowire can be extracted from the slope of this linear relationship. A single GaN nanowire, with 2 µm length, 9.8×10-10 cm2 hexagonal cross-sectional area and 1019 cm-3 nominal dopant concentration, was detached and lifted up from its substrate in-situ in a scanning electron microscope (SEM) (Fig. 3. (1a)), and placed on two pre-patterned W plates (Fig. 3. (1b)). Off-axis electron holograms were acquired as a function of biasing current. The phase image obtained without biasing was used as a reference for the measurement of the phase change along the nanowire. The contribution of electric field leakage to the vacuum, from the electrical contacts between the nanowire and W plates to the phase shift, was estimated from the phase profiles in the vacuum (arrow B in Fig. 3. (2b)) and subtracted from the phase profile along the nanowire at each current. The resistivity of the nanowire was measured to be (8±1)×10-4 Ω-cm.



Figure 3. (1a) Detaching a single nanowire from its substrate in an SEM using a micromanipulator and placing it on a pre-patterned SiN membrane. (1b) SEM image showing a 2 µm nanowire sitting on a 1 µm gap in between two W plates. The gap was micromachined on a W lamella using focused ion beam milling. The SiN membrane was partially milled away to provide the required vacuum region for electron holography. (2a) Hologram acquired from an electrically connected GaN nanowire. The nanowire thickness was measured directly from the hexagonal geometry of nanowire. (2b) Reconstructed phase image. At each biasing current, the phase profile in a vacuum (arrow B) was subtracted from the phase profile along the nanowire (arrow A), to compensate for the leakage of electric fields from the electrical contacts to the vacuum. Arrow I shows the direction of the applied current. (2c) Corrected phase profiles as a function of biasing current. It can be seen that by increasing the current the slope of the phase profile increases. (2d) Phase differences measured along 800 nm of the nanowire as a function of current. From the slope of the fitted line in this graph, the resistivity of the nanowire was measured to be (8±1)×10-4 Ω-cm.

Connection between misfit dislocations and stacking faults in InN dots revealed by focused ion beam milling InN quantum dots were grown on GaN wires by MOVPE. Figure 4 shows the SEM images of InN dots which were grown for different times on GaN wires. The images show that InN dots tend to form at the wire edges between side facets. Also, the InN dots do not form until 30 s after initiating growth. With increasing the growth time from 30 s to 180 s, the dot height increases from 4.9 nm to 19.9 nm.



Figure 4. InN quantum dots grown for different times on GaN wires. Both top view and side view SEM images are shown here and the growth time is marked also. The GaN wires were pulled down onto the substrate surface to take side-view images. The GaN wires are about 2 um long. Scale bar in top view SEM images: 200 nm. TEM images of InN dots on GaN cores reveal misfit dislocations in the <1-10> viewing direction, and stacking faults in the <11-20> viewing direction. Unfortunately the projected thickness through the nanowire is too large in <11-20> to simultaneously see the misfit dislocations. To circumvent this we utilized focused ion beam (FIB) milling to cut the nanowires in half (illustrated in the inset in figure 5c). To ensure that the small dots were not damaged by the ion beam we developed a technique where the nanowire array was first embedded in a protective resist layer.



Figure 5. a Overview of the FIB prepared TEM sample with over 30 thinned nanowires. An SEM image of the same sample is shown in b. A high resolution TEM image is shown in c, where the nanowire has been sufficiently thinned to clearly reveal the stacking of the c-planes. Here the stacking faults can be seen to originate in misfit dislocations at the dot-nanowire interface, which was only hinted at in the images from the un-thinned nanowires in figures 1-3.

“Report on TEM and FIB based microscopy studies …...phase shift, was estimated from the phase...

Documents

Transcript of “Report on TEM and FIB based microscopy studies …...phase shift, was estimated from the phase...