In Situ and Analytical Transmission Electron Microscopy
Mechanism of Layer Exchange in a-Si/Al Thin Film Structures
- Light microscopy image (reflection mode) of a 10% ex situ reacted plan-view TEM sample as viewed from the quartz substrate side. Primary Si grains, secondary Si grain and, secondary Si nuclei free denuded zone around primary Si grain are indicated.
- Fig. 1: Schematic diagram of the Al induced layer exchange (ALILE) and crystallization process.
- Fig. 2: (a) Al Kα STEM-EDXS elemental map of the region marked by box in the eye catcher image. The schematics of cross-section geometries corresponding to the various regions marked in STEM-EDXS elemental map are shown on the right. (b) Plan-view TEM bright-field image of pushed-up Al in the 60% reacted sample. (c) Schematic view of islands of pushed-up Al on top of Al grains and grain boundaries in the non-reacted part of the 60% sample.
- Fig. 3: (a) TEM bright-field image of the non-reacted region in a 60% ex situ reacted sample in cross-section and (b) the corresponding RGB overlay of the Al-K (red), Si-K (green) and O-K (blue) STEM-EDXS elemental maps. The two selected area diffraction patterns confirm the epitaxial relationship between the pushed-up Al (region “1”) and the underlying Al-grain (region “2”). (c) HRTEM image close to the a-Si/Al interface. Discontinuous amorphous Al oxide layer at the a-Si/Al interface is marked. The (200) atomic planes (indicated by dashed line) are continuous across the pushed up Al / Al interface.
- Fig. 4: Plan-view TEM bright-field snap shot images of the advancing c-Si growth front of the dendritic cells like the one shown in figure 2 (a) during in situ heating. Sudden push-up of Al is revealed in (a-b) while grain boundary realignment is revealed in (c-d). The relative time at which the images were acquired is indicated.
In situ and analytical transmission electron microscopy (TEM) has been used to investigate the mechanism of material transport during Al-induced layer exchange (ALILE) and crystallization of amorphous Si (a-Si). Significant lateral and vertical redistribution of Al was observed, yielding Al deficient dendritic cell centers surrounded by an about 10 µm wide Al excess zone containing epitaxial islands of "pushed-up" Al whose number density and size decreases with increasing distance from the cell boundary.
Transmission electron microscopy is unique in its ability to simultaneously provide bulk structural and chemical information of a thin (<100 nm) sample, with sub-nm lateral resolution. This can primarily be attributed to the high energy electron beam used as a probe in the TEM. A typical 300kV electron beam has an extremely small (~2 pm) wavelength which enables sub-nm lateral resolution. Strong interaction of the primary electron beam directly with the charged particles (electron and nuclei) in the matter yield a number of signals e.g. characteristic x-rays, inelastically scattered electrons with energy loss, electron elastically scattered by the nuclei at high scattering angles, etc. These signals give rise to analytical techniques e.g. energy dispersive x-ray spectroscopy (EDXS) and mapping, electrons energy loss spectroscopy (EELS) and energy-filtered transmission electrons microscopy (EFTEM), as well as Z-contrast imaging.
In situ TEM
A working description of in situ TEM by F. M. Ross  can be summarized as follows: the class of experiments where a specimen is acted upon by a controlled change to its environment (input) and the resultant change (output) in the specimen is measured while the specimen is still inside the TEM. The input parameters can be temperature, gas pressure, strain, electric field, magnetic field, etc. The output can be diffraction contrast images, high-resolution (HR) TEM images, diffraction patterns, EDXS signal, EELS signal, Z-contrast images, etc. In an in situ TEM, a physical phenomenon is tracked continuously, catching any transient phases or nucleation events etc which enables one to accurately understand the mechanism of physical processes.
Layer Exchange and Crystallization in a-Si/metal Bilayers
When a-Si (a-Ge)/metal (eg Al, Ag) bilayer is heated, semiconductors and metal layers exchange their position, but the Si appears in a crystalline phase in its new layer position (fig 1).
This metal induced amorphous to crystalline phase transformation occurs at temperatures much lower than the solid phase crystallization temperature and enables the use of cheap substrates like glass or even polymers. Secondly the size of Si grains can be as large as 100 µm when a suitable oxide or Ti interface layer is placed between the a-Si and metal layer. The production of such large grain size Si films on cheap substrates has attractive commercial applications in photovoltaic (solar cells) and optoelectronic (thin film transistors) industry.
In situ TEM is extremely useful in the study of phase transformation in materials. In particular, metal induced crystallization (MIC) of a-Si and a-Ge has been studied by in situ TEM by Herd et al. , Konno and Sinclair , and recently by the authors . The current understanding of the MIC and layer exchange process in the a-Si/Al system can be summarized as follows: (i) dissolution of Si from the a-Si top layer into the Al bottom layer, (ii) nucleation of Si crystallites at grain boundaries of the Al layer, (iii) lateral growth of dendritic Si crystals in the original Al layer and (iv) transport of Al into the a-Si layer. A number of questions regarding the MIC are yet to be answered: (i) the mechanism of Si transport in the Al layer, (ii) mechanism of Al transport, (iii) the role of oxide interfacial layer. The last two questions were investigated by a combination of in situ and analytical TEM by the authors in [4, 5] which will be summarized here.
Bilayers of a-Si(~100nm)/Al(~50nm) on quartz glass substrates were prepared using thermal evaporation and electron-beam evaporation for Al and a-Si deposition, respectively. The ALILE and crystallization reaction was studied by bright-field video imaging (27 frames / min) during annealing of a plan-view sample at 410°C in a Philips CM30 TEM operated at 300 kV. Additionally samples annealed ex situ at 450°C were investigated by scanning (S)TEM-EDXS mapping and HRTEM imaging using TITAN3 80-300 (FEI) at 300 kV equipped with an image side aberration-corrector (CEOS) and an EDXS detector (EDAX). Plan-view and cross-section TEM samples were prepared by conventional method comprising mechanical thinning, dimpling, followed by 3 kV Ar+ ion milling.
Results and Discussion
A reflection light microscopy image of a typical a-Si/Al/quartz stack annealed ex situ to 10% layer exchange is depicted in the eye catcher figure. The image was taken from a plan-view TEM sample in order to allow for direct correlation with TEM data (see below). In the light microscopy image primary crystalline Si grains appear dark and dendritic. In addition, smaller secondary Si nuclei formed upon cooling are also present. The sizes of the primary Si grains in the samples range between 5 µm and 50 µm.
An Al-Kα STEM-EDXS elemental map around a 6 µm large crystalline Si grain marked by a box in the eye catcher figure is shown in figure 2 (a). For various regions marked in figure 2 (a), the vertical stacking of Al and Si is schematically illustrated on the right of the image. The non-reacted region, marked by the topmost box, has the geometry of the starting stack. The box below has the darkest contrast and is therefore almost devoid of Al. It indicates that this region consists of Si in the top layer, as well as in the bottom layer which initially consisted of Al. The third box from the top depicts a region consisting of crystalline Si in the bottom layer and a mixture of Si and nanocrystalline Al in the top layer. In total it can be concluded that the region comprising this early stage Si grain primarily contains Si and is thus severely depleted of Al. The lowest box, which like the topmost box lies in the non-reacted region, depicts an area consisting of Al layer in the bottom layer and Al in the top layer, which we named "pushed up" Al . The density and size of such pushed-up Al islands decreases with increasing distance from the reaction front. Such a pushed-up Al/Al geometry with an epitaxial relationship was confirmed by diffraction contrast microscopy in plan-view (fig. 2(b)) and cross-section geometry (fig. 3 (a, b) and by HRTEM in cross-section geometry (fig. 3 (c)) [4,5].
A direct evidence for the Al push-up in the non-reacted regions was obtained by in-situ heating of the a-Si/Al/quartz stack at 410 °C inside the TEM . Snap-shots from the videos acquired during this in-situ experiment are shown in figure 2 (see  for complete videos). The annealing time at which these snap-shot images were acquired are indicated. In comparison to figure 4 (a), figure 4 (b) (2 seconds later) showed sudden appearance of pushed-up Al (indicated by black arrow) at a considerable distance from the reaction front. Another frequently observed phenomenon during in situ annealing is the significant Al grain boundary realignment in the proximity of the expanding reaction front, as shown for example for grain 7 in figure 4 (c, d).
Summary and Outlook
The important findings [4, 5] revealed by in situ and analytical TEM are: (i) formation of Al-deficient centers in the reacted regions, (ii) excessive upward transport of Al in the form of pushed up Al islands in a ~10 µm wide rim of non-reacted material around the dendritic cell, (iii) epitaxial relationship between the islands of pushed up Al and the underlying Al grains, (iv) decrease of the density of pushed up Al islands with increasing distance from the cell boundary, (v) significant realignment of Al grain boundaries close to the growing reaction front and (vi) indication of stress supporting push-up of Al [4, 5]. This work demonstrates the capability and usefulness of combining in situ and analytical techniques available in modern TEMs in understanding the precise mechanism of physical processes in materials.
Tobias Antesberger and Prof. Martin Stutzmann from Walter Schottky Institute are gratefully acknowledged for providing the Alile samples.
 Ross F. M.: in Science of Microscopy, ed. P.W. Hawkes, J. C. H. Spence, Springer, 2008, 445-534
 Herd S.R. et al.: Non-cryst. Solids 7, 309 (1972)
 Konno T.J. and Sinclair R.: Philos. Mag. B 66, 49 (1994)
 Birajdar B. et al.: Scripta Materialia 66, 550 (2012)
 Birajdar B. et al.: Phys. Status Solidi RRL 5, 172 (2011)
Dr. Balaji Birajdar
Prof. Dr. Erdmann Spiecker (corresponding author)
University of Erlangen-Nürnberg
Center for Nanoanalysis and Electron Microscopy
Materials Science Department