HIGH-RESOLUTION ELECTRON-BEAM INDUCED CURRENT ...

23rd European Photovoltaic Solar Energy Conference, 1-5 September 2008, Valencia, Spain

HIGH-RESOLUTION ELECTRON-BEAM INDUCED CURRENT IMAGING OF THE P-N JUNCTION IN CRYSTALLINE SILICON ON GLASS (CSG) SOLAR CELLS Martina Werner1, Christian Hagendorf2, Otwin Breitenstein3, Frank Altmann2 and Jörg Bagdahn1 1 Fraunhofer Center for Si-Photovoltaics, 2Fraunhofer Institute for Mechanics of Materials Walter-Hülse-Str.1, 06120 Halle (Saale), Germany 3 Max Planck Institute of Microstructure Physics, Weinberg 2, D-06120 Halle (Saale), Germany

ABSTRACT: The p-n junction of crystalline silicon thin film solar cells on glass (CSG material) has been studied in correlation to high-resolution microstructure analysis. Scanning and transmission electron microscopy (SEM/TEM) have been used to study material-related properties like texture, grain structure and layer structure. Complementary information on the p-n junction properties was obtained by high-resolution electron beam induced current (EBIC) imaging. Beside a plan-view EBIC investigation of the surface, a beveled cross sectional preparation has been applied to study the location of the p-n junction in relation to grain structure and thin film topography. The depth of the p-n junction was roughly estimated to less than 150 nm. The crystal and the layer structure of the films have been studied by TEM in greater detail. Keywords: EBIC, electron microscopy, silicon solar cells, thin films

1

silicon is etched through to the n+ layer. The second stage produces the positive polarity dimple contacts to the p+ layer. Aluminum is deposited over the entire rear surface of the device and patterned into thin stripes joining the n+ to the p+ region.

INTRODUCTION

Thin film technology for silicon solar cell production is a promising alternative in terms of a reduction of both cost and material. The application of thin silicon layers on economical glass substrates in combination with a defect tolerant interconnection scheme has lead to a quick commercialization of Crystalline Silicon on Glass (CSG)-modules with record efficiencies above 10% [1, 2]. The performance of thin film solar cells is strongly determined by a profound understanding and control of the interaction between microstructure and local electrical properties. The processing of CSG solar cells comprises the deposition and crystallization of the Si film as well as the fabrication of a fault tolerant interconnection scheme. Due to the complex production process a number of process- as well as material-induced electric defects (e. g. p-n junction formation, parallel and serial resistances) can be determined. Work has been successfully spent into an optimization of the interconnection scheme using for example methods like IR imaging, lock-in thermography and potential mapping [3, 4]. In this work the focus is on the study of materialrelated electrical properties in correlation with microstructure analysis. In particular, high-resolution cross-sectional electron-beam induced current (EBIC)imaging is applied to study the p-n junction in correlation with microstructure material properties. Beside a planview EBIC investigation of the surface, beveled crosssectional preparation has been applied to study the location of the p-n junction in relation to topography and grain structure. The crystal and the layer structure of the films have been studied using TEM in greater detail. A short introduction to the layout scheme of the CSG thin film cells may be helpful in order to understand the micro-structural data presented. A scheme of the CSG device is shown in Fig. 1. A textured glass is the substrate, on which a thin silicon nitride layer is deposited followed by three other layers of differently doped silicon. The silicon is patterned by forming grooves using a laser. A few microns of an optically nonabsorbing but electrically insulating resin is then applied and patterned in two stages. The negative polarity crater contact regions are produced in one stage, where the

Figure 1: CSG device structure [2] 2

EXPERIMENTAL SETUP

The laterally separated arrangement of craters and dimples in the CSG module allows a simple preparation of the samples for planar Scanning Electron Microscopy (SEM) and EBIC investigations. The preparation for planar EBIC measurements included the removing of resin and Al layer in an ultrasonic bath of acetone for three minutes. The crater and dimple which still contain the Al layer were directly contacted by needles. Different preparation methods have been tested to prepare cross sections suitable for high-resolution cross sectional EBIC investigation of the p-n junction. Firstly, samples were broken at a defined location. Furthermore, ion polishing was applied in JEOL SM-09010 Cross Section Polisher. Finally, a beveled cross section was prepared with water on a glass sheet in an angle of 6°. For this preparation procedure the resin had to be removed. Scanning Electron Microscopy and EBIC investigations were performed in a ZEISS SEM Supra 55VP. This microscope was completed by a Specimen Current-EBIC (Electron Beam Induced Current) Amplifier of the company K&E Developments. For target preparation a Focused Ion Beam (FIB) system was used, particularly to prepare the TEM lamellas which were investigated by a CM20 from Philips. 2217


3

Fig. 2b shows the EBIC image of the same region. The investigations were performed with 3 kV primary electron beam energy. The beveled edge of the crater is easily visible in this image as well as in the SE image (arrows). A clear assignment of the position of the p-n junction from this image is difficult, since the EBIC contrast is obtained from an extended region. Furthermore, the geometry of the crater edge is not well known, because of the etching procedure. Thus, the depth of the p-n junction cannot be determined from its lateral position in the image. The jagged appearance may be the result of a superposition with the surface topography as well as the Si grain structure. On top of the Si film surface strong variations of the EBIC contrast are observed in flat regions (F) as well as on the beads (B). This behavior is even more pronounced in Fig. 2d and marked by circles. Fig. 2c and d were measured with 10 kV primary electron beam energy. Fig 2c shows the SE image with Si bead structures. In the flat area between the beads (F) a homogenous EBIC signal should be observed as expected from a flat and homogenous p-n junction. However, slight variations in the EBIC contrast are visible, which could be caused by different properties of individual Si grains. Up to now it is not clear whether the grain structure, stress or any other properties influences the EBIC intensity. The Si bead structures themselves show pronounced differences in the EBIC signal (B). Some Si bead structures have a low EBIC collection efficiency although the topographical structure quite identical and the p-n junction should follow the contour of the Si film surface on the beads. Acquisition of EBIC images at different electron energy from 2-20 kV result in strong contrast changes as can be seen by comparing Fig. 2b and d. By increasing the acceleration voltage of the electron beam a more intense EBIC signal is obtained from surface regions in comparison to the p-n junction at the crater edge (not shown here). This can be explained by the increasing penetration depth at higher electron energies. The energy dependent average penetration depth has been calculated for Si using a Monte-Carlo simulation software (Casino) [5]. Values are given in Table 1. They coincide with commonly cited models [6]. Acceleration voltages of 3 kV correlate with a penetration depth of 100 nm. Thus, it can be assumed that at low voltages (< 5 kV) less electrons reach the p-n junction and a reduced EBIC signal at the surface is observed. Increasing the energy above 5 kV yields a strong EBIC intensity. Thus it is concluded, that an effective electron hole separation within the built-in field of the p-n junction occurs in a depth of 100 – 300 nm. With reference to the complete Si layer thickness of 1.5 µm the p-n junction must be located close to the Si surface.

RESULTS AND DISCUSSION

3.1 Plan-view EBIC imaging A top view SEM image of a CSG thin film surface after removal of the Al coating and resin shows the characteristic morphology related to the SiO2 bead texture (Fig. 2a). Small beads (diameter of 500 nm) are positioned underneath the silicon and produce this typical structure of the silicon surface. In the left part of Fig. 2a a crater region is visible where the Al is still present on the etched Si film. The crater edge can be clearly identified as a sharp line (arrows). The dark holes in the crater are pin holes (PH), which are produced to optimize the contact to the n+-layer. Due to the pin holes the Al contacted area is increased, resulting in a reduced contact resistance. a

P B F

b

B F

d

c F

B

F

Table1: Correlation between acceleration voltage and average penetration depth of electrons in Si [5]

B

Acceleration voltage [kV] 2 3 5 10 15 20

5 µm Figure 2: SE image of a crater edge (a), EBIC image of the same region showing the p-n junction at 3kV(b), SE image of the Si bead structure, primary electron beam energy 10 kV (c) EBIC image of the same region at 10 kV (d) 2218

Penetration depth [nm] 40 100 300 500 1300 2000


A detailed understanding of the observed plan-view EBIC contrast is still difficult, because of the complexity of the involved parameters like recombination activity and diffusion length. Topographic features may also play an important role [7], which is particularly important for the strong contrast changes at the Si bead structures.

c

3.2 Cross-sectional EBIC imaging A more direct access to the Si layer structure and the location of the p-n junction was expected from EBIC at cross sections samples. In the first step cross sections were prepared by breaking the solar cells at defined locations. As a result the location of the p-n junction could be investigated close to craters and dimples (not shown here), but the jagged broken surface did not allow to distinguish between topography, grain structure and inhomogeneities of the p-n junction. In a second step ion polishing was applied for cross section preparation. It turned out that by the ion bombardment the electronic properties are strongly disturbed. The damage of the ion beam probably leads to a higher surface recombination and a weaker EBIC signal of the p-n junction. Higher acceleration voltage was required to image the p-n junction. But the resolution was reduced at higher magnifications so that details could not be resolved. Finally, a mechanical beveled polish (6°) with water on a glass sheet was performed. The SE and EBIC images are shown in Fig. 3. In the SE image (Fig. 3a) four horizontal regions can be distinguished: A) the area at the upper image edge contain the unpolished material. Bead structures and Si layer are not affected.

a

Figure 3: SE image of a beveled polish (a), EBIC image of the same region (b), superposition of SE and EBIC images (c) B) Below this region polishing tracks at the surface of the silicon covering beads are visible. C) The region below shows beads which are polished followed by D) completely polished surface. The grain structure (G) of the multi-crystalline silicon is clearly visible. The average grain size is 2 µm. All EBIC images of the beveled polish were taken with 2 kV which corresponds to a penetration depth of 40 nm. In the EBIC image (Fig. 3b) the p-n junction is clearly detected and shows a bright contrast with a width of 150 nm at half maximum. This contrast only appears at locations where the bevel cuts through the p-n junction. Due to the small penetration depth of 40 nm only few electrons reach the p-n junction through the surface in the unpolished regions. The location of the p-n junction is strongly influenced by the topography of the bead structures. For better demonstration a scheme is shown in Fig. 4.

A

p-n junction

B beads

C

D

silicon

glass

G

b

Fig. 4 Cross section scheme of the surface (a) and beveled polish (b) In the unpolished region A the EBIC contrast is similar as in Fig. 2b. Due to the low acceleration voltage of 2 kV, also here the cut p-n junction shows a much higher EBIC signal than the unpolished region. Circles in Fig. 3b mark areas with different EBIC contrast. The small yellow circle shows a flat surface between the beads with the typical Si contrast. At the small green circle no EBIC contrast is observable. The SE image (Fig.3a) revealed polishing residua, which shaded the underlying p-n junction. The red circle shows two impressions without any EBIC contrast. These are regions where the p+- region is isolated from that of the

2219


4

whole sample by polishing process, so that it is not connected with the EBIC amplifier. From the SE image and the size of the beads (TEM analysis), the depth of the p-n junction can be roughly estimated to less than 150 nm. Referring to the intended CSG layer model (Fig. 1) with an over all layer thickness of 1.5 µm, this corresponds clearly to the uppermost junction. At the location of the intended junction (Fig. 1) no EBIC signal was observed. It seems that in the present case the intended p-layer was actually an n-layer, so the maximum EBIC signal is produced on the junction of the p+ and n layer which is located in the upper area of the silicon. Similar results were also reported by D. Inns et al [8]. They investigated poly-silicon thin-film diodes on glass by high resolution EBIC and found the p-n junction as a p+-n junction.

CSG solar cells were investigated using SEM, EBIC and TEM with focus on the study of material-related electrical properties in correlation with microstructure analysis. Planar and cross sectional EBIC measurements were performed to investigate the location of the p-n junction. Different preparation methods have been tested to prepare cross sections suitable for high-resolution cross sectional EBIC investigation. Finally, a beveled cross section prepared with water on a glass sheet in an angle of 6° was successfully applied. The p-n junction was found in a depth below 150 nm following the SiO2 bead structure. Planar EBIC images revealed variations of the contrast in flat regions. The origin of this behavior may be possibly related to the grain structure of the film. TEM studies of a crater bottom containing pin holes showed that this region is defect free with homogenous layers of Si3N4, Si and Al. With respect to the shallow depth of the p-n junction it can be assumed that the crater bottom is located completely below the p-n junction. The pin hole formation reduces the resistance by increasing the contact area between Al metal and the n+-Si-layer. The performed investigations contribute to a better knowledge of the p-n junction in CSG devices and can be a step toward further studies of material-related electrical properties in correlation with micro-structure analysis.

3.3 Microstructure analysis by TEM TEM sample preparation was performed using FIB technique. The goal was the analysis of the microstructure of a crater bottom region containing pin holes. In the TEM micrograph in Fig.5 such a region with a pin hole is shown. On the glass substrate four beads are visible with an average size of 500 nm. A thin Si3N4 layer covers the beads. The silicon layer above shows clearly the poly-crystalline structure. Usually the Si layer is 1.5 µm thick but in this case it is thinner because the crater is etched into the silicon. Above the Si layer the Al layer is visible. The pin hole reaches through the Si layer to the Si3N4 layer.

ACKNOWLEDGEMENTS This work has been supported by BMBF project SiThinSolar (03IP607). Support by J. Schneider, CSG Solar is gratefully acknowledged.

7 aluminum

Si3N4

SUMMARY

REFERENCES

Pt protection [1] M.A. Green et al, Solar Energy 77 (2004) 857-863 [2] M.J. Keevers et al, Proceedings 22nd European Photovoltaic Solar energy Conference, Milan (2007) [3] A. Straub et al, http://www.csgsolar.com/downloads/ CSG_Paper_Straub_et_al_-_3AV.1.11_preprint.pdf [4] O. Breitenstein et al, J. Appl. Phys. 102, 024511, 2007 [5] Casino Software, D. Drouin et al, Scanning 29 (2007) 92-101 [6] Kanaya and S. Okayama, J. Appl. Phys. D: Appl. Phys.5 (1972) 43, T.E Everhart and P.H. Hoff, J. Appl. Phys. 42 (1971) 5837 [7] Sproul et al. Proceedings 2nd World Conference of Photovoltaic Solar Energy Conversion, Vienna, Austria, July 1998, p1355 [8] D. Inns. T. Puzzer, A.G. Aberle Thin Solid Films 515 (2007) 3806-3809

silicon pin hole bead voids glass

Figure 5: TEM micrograph of the crater bottom region The micro-structural data obtained from TEM analysis confirms parameter of the layer structure. The average bead size is 500 nm in diameter. The grains size as well as the size of the pinholes is 0.5 - 2 µm. Voids are visible underneath the beads due to shadowing during the deposition process. The Al layer covers the crater bottom including the pinholes homogeneously, resulting in defect free contacts with low resistance. Whereas the dimple bottom is on the top of the p-n junction the crater bottom is located completely below it.

2220