Abstract
The visual appeal is an important criterion for the acceptance of photovoltaic modules in facades and roofs or buildings, or in village electrification. Therefore, on the one hand, we design artistic front contact patterns converting the busbars into an attractive feature without an undue decrease of efficiency. Such designs can be applied on mono- as well as on multicrystalline silicon wafers. We show that our recently designed patterns lead to lower efficiency, due to increasing resistivity and shading losses, of only about 0.7% absolute compared to the standard pattern under 1-sun conditions. On the other hand we develop a method based on pattern recognition to apply the finger grid on the front side along the grain boundaries of the multicrystalline silicon. This will increase the area of high-efficiency inner grain regions and reduce series resistance losses at the grain boundaries. Therefore an overall increase of efficiency can be expected.

Introduction
The acceptance of photovoltaic modules in highly visible places like walls and roofs of buildings is to great extent determined by non-technical aspects, most of all by the visual appeal. The design of the surfaces of the cells and modules must therefore meet two optimisation criteria: High energetic output and attractive appearance. Both aims can be achieved in many different ways, for example by differently coloured cells [1]. In the present work we focus on the front contact metallisation of multicrystalline silicon solar cells.
The first part of our work deals with the design of new artistic front contact patterns instead of the standard pattern, often called the "H". As the busbars of the front collection grid are often considered visually annoying, we have tried to incorporate artistic shapes into them. We have chosen the busbars because they are lines of sufficient thickness to be noticeable, even from the distance. We have created fifteen new patterns and modified the fine finger grid where necessary. We have investigated the resistive losses in the emitter layer, resistive and contact losses in and at the fingers, resistive losses in the busbars and losses due to the shading by fingers and busbars. Assuming an illumination of 1000 W/m² (1-sun), a current generation density of the silicon cells of 31.3 mA/cm² and a maximum-power-point voltage of 503 mV we have found out that a 100 x 100 mm² solar cell with the standard busbar pattern would exhibit an efficiency of 13.80%. Under the same conditions our cell with the poorest electrical performance would show an efficiency of 13.09% due to the non-ideal metallisation pattern.
The second part of this work deals with the increase of efficiency due to a metallisation along the grain boundaries. Investigations of the charge transport across grain boundaries in multicrystalline silicon solar cells [2], [3] showed the possibility of reducing the average sheet resistance across the grain boundaries in the highly doped emitter of multicrystalline solar cells by about 30 percent by contacting along the grain boundaries. As a consequence one can find an increase of the curve fill factor. H.W. Kim et al. suggest in [4] a combination of the grain boundary contacting method and a conventional grid electrode to minimise the effect of grain boundary barriers. Our aim is to set up a production process starting with a multicrystalline silicon wafer of 100 x 100 mm² size. After the digitisation by a flatbed scanner the grain boundaries are detected by a computer program. A command file is created that forces an xy-table to "paint" the metallisation grid onto the grain boundaries.

Artistic patterns and modules
Our basic concept is "to see the module, not just the cell". The bus bars were thus designed with the intention to permit a wide variety of overall bus bar patterns of the module with just a few elementary cell patterns. We present fifteen metallisation patterns for the front side of crystalline solar cells. Some patterns allow to create any combination of big size quadratic and rectangular panels. Other ones are an attempt to break out of the "two bus bars"-principle. They have connection points in the middle of the four sides of the wafer, but the inner layout of the bus bar is fully asymmetric (Patterns 1 (Crack) and 2 (Fissure) in Fig. 1). This permits a tremendous number of ordered as well as chaotic patterns in a panel with just a random combination of only two cell types (Fig. 1)! The patterns 3,4 and 5 in Fig. 2 are based on a visual one third - two thirds division of the squared wafer and allow to create a variety of art deco features, an example of which is shown as a module of 6 rows and 11 columns in Fig. 2. Clearly, electrical series connection between cells in a module cannot always be as the bus bar pattern suggests. Sometimes leads will have to be isolated and run underneath a few cells to the next connection. The fingers have a spacing of 2.941 mm from centre to centre and an assumed width of 120 µm. The width of the busbars is 2 mm. This spacing is more or less optimal for the chosen sheet resistance of the emitter layer and has been kept for all the designs. The series resistance in the busbars is sufficiently small and has therefore been neglected. Our calculations presented in [5] show that for a cell with an active area of 100 x 100 mm² the additional loss in efficiency is essentially due to increased shading by the new patterns and to increased series resistance of the fingers.

Fig. 1: Module "Crack/Fissure" and the corresponding single cell patterns "Crack" and "Fissure"

Fig. 2 Module "ArtDeco" with the corresponding patterns "A_deco", "B_deco" and "C_deco"

Table I gives the calculated power losses and efficiencies under 1-sun conditions of the five patterns presented in Fig. 1 and 2 compared to the standard pattern:

Pattern	Electrical loss (finger, sheet, contact) [mW]	Total shading loss [mW]	Total loss [mW]	AM 1 Efficiency [%]
Standard	69,8	124,7	194,5	13,80
Crack	97,1	148,6	245,7±4,4	13,29±0,04
Fissure	101,5	130,2	231,7±13,5	13,43±0,13
A_deco	60,1	171,7	233,9±15,0	13,41±0,15
B_deco	90,2	123,8	214,2±10,1	13,60±0,10
C_deco	103,1	162,1	265,5±0,2	13,09±0,00

The electrical losses are given by the sum of losses due to the emitter sheet resistivity, emitter - finger contact resistance and finger series resistance. Adding the shading losses of both the fingers and the busbars leads to the values of the "Total shading loss". The uncertainties in the column "Total loss" result from the estimated inaccuracies in evaluating the electron collection by the fingers in the analytically cumbersome patterns.
Although the familiar "Standard" busbar pattern gives higher efficiency than any of our new ones the efficiency will only drop about 0.7% absolute for the worst-performing pattern, assuming 1-sun conditions. This seems a small loss considering the enhanced architectural value. The loss will be even less under typical illumination conditions of visible facades and roofs.

Grain boundary effects
It is a well known effect that grain boundaries in multicrystalline silicon solar cells degrade the solar cell performance. Several publications, for example [6], deal with the various influence factors of the losses (recombination, resistive effects) at the grain boundaries. It has been shown that printing the metallisation contact of the front surface of the solar cell directly onto the boundaries of two adjacent grains improves the solar cell characteristics [2], [3], [7]. Due to the metallisation of the back side and the front contact grid the individual grains are in a parallel circuit. Usually the front side grid is not made in such a manner that every grain has its own metal contact. As a consequence the current generated by the incident light within a grain has to pass through grain boundaries which are regions of enhanced electrical resistivity . These electrically active grain boundaries due to the formation of a potential barrier at the boundary plane contribute to the overall series resistance of the solar cell and can for example be detected by high resolution topographies of the specific resistivity [7]. Since grain boundaries without passivation are mostly areas of enhanced recombination for the light generated minority charge carriers the internal quantum efficiency is reduced near the grain boundaries whereas regions within the grain (with a high quantum efficiency) may be shaded by the front contact metallisation grid and thus cannot contribute to the generation of electron-hole pairs. The basic idea of the present work is to print the front contact grid along the grain boundaries wherever possible and to show the improvement compared to a conventional metal grid.

Production line
The starting point of the production of grain-boundary contacted multicrystalline silicon solar cells is an optical digitisation of the surface of the wafer with a flatbed scanner. The scanning resolution is 1016 dpi in both the x and y direction to give a one-point-size of 25 µm and the colour depth is 8 bit (256 values). The other input is a rectangular grid of finger lines (the "net"). In a computer program this net is compared to the wafer image and distorted such that it follows the grain boundaries as much as possible, the restriction being that the distance between any point on the wafer to the nearest finger line must not exceed a certain maximum. This latter criterion is set according to the sheet resistivity of the emitter layer. The calculated grid pattern is then translated into a code to control an xy-table to draw the lines directly onto the wafer. The printing process is currently done by galvanisation of silver nitride with an appropriately adapted pen. We are also investigating the options of drawing the lines with conventional screen printing paste, and of defining them lithographically with subsequent electroless plating.

Result

Fig. 3: Grain boundary pattern (red lines) on a piece of a multicrystalline silicon solar cell.

Fig. 3 shows a 24,8 x 24,8 mm sector of multicrystalline silicon solar cell with its grain structure and an added "net" grid calculated in the way mentioned above. Further investigations are in progress concerning the printing of the metallisation grid. Different tips and different printing velocities are tested to optimise the separation of the silver. Measurement of the resistivity of the grid will be made and presented.

Conclusion
We presented new metallisation patterns for multicrystalline silicon solar cells with an improved visual appeal and still enough power generation. Despite the fact of new design requirements such as only one busbar contact per side we calculated an overall efficiency loss of about 0.7 % absolute under 1-sun conditions in the worst grid case. We also showed a new production process of the metallisation of the front side of multicrystalline solar cells based on grain boundary recognition and silver galvanisation. The process is at the moment under completion but working in principle. It might come out that it could still be necessary to add busbars to the grain boundary grid to assure the transfer of the generated current to the connection tabs. We assume to be able to present resistivity results in the nearer future.

Acknowledgement
Part of this work is supported by a European grant under the Joule program Nr. JOR3CT970175.

References

[1]	N.B. Mason, T.M. Bruton, R. Russel, R.D.W. Scott, The Development of coloured silicon solar cells for architectural applications, presented at Eurosun '96, Freiburg, Germany, 1996.
[2]	J. Summhammer, V. Schlosser, Investigations of a novel front contact grid on poly silicon solar cells, Proc. 12th European PVSEC, Amsterdam, The Netherlands, 1994, 734 - 736.
[3]	V. Schlosser, J. Summhammer, A study of the charge transport across grain boundaries in multicrystalline silicon solar cells with a front contact along the grain boundaries, Proc. 14th European PVSEC, Barcelona, Spain, 1997, 716 - 719.
[4]	H.W. Kim, D.G. Lim, S.E. Lee, S.S. Kim, J. Yi, Poly-Si solar cells with an electrode along the grain boundaries, presented at 2nd World PVSEC, Vienna, Austria, 1998.
[5]	M. Radike, J. Summhammer, Electrical and shading power losses of decorative front contact patterns, in refereeing process for "Progress in Photovoltaics".
[6]	M.A.: Green, Silicon Solar Cells: Advanced Principles & Practice, Centre for PV Devices, UNSW, Sydney, Australia, 1995.
[7]	C. Häßler, W. Koch, W. Krumbe, S. Thurm, A. Müller, I.A. Schwirtlich, Multicrystalline BAYSIXÒ silicon for high-efficient solar cells from the new Freiberg production facility, presented at 2nd World PVSEC, Vienna, Austria, 1998.

[BACK] [MAIN] [FURTHER]