A. Kress (Screen printed Emitter-Wrap-Through back contact solar cells) B. Terheiden (Lamella-Silicon solar cells)
A. Boueke and R. Kühn (Latest results on semitransparent POWER silicon solar cells)
S. Keller, S. Scheibenstock and M. Wagner (Theoretical and experimental behaviour of monolithically integrated crystalline silicon solar cells based on the HighVo solar cell concept)
W. Jooss and K. Blaschek (Back contact buried contact solar cells) P. Fath and E. Bucher
Screen printed Emitter-Wrap-Through (EWT) back contact solar cells
Unlike conventional cells, back contact solar cells have both electrodes on the rear side. This avoids shad-owing losses (summing up to 12 % for conventional cells) and facilitates considerably the connection of cells to modules. In addition the rear side emitter contributes to a higher collection probability of minority carriers generated near the cell rear. The EWT-concept is spe-cially suited for low cost silicon with diffusion lengths smaller than the device thickness. Small laser (Nd-YAG) drilled holes connect the front side emitter with the emitter contact at the cell rear. Screen printing, a re-liable and cost efficient technology in industrial practise, is applied for contact formation. Different geometryies (fingerlength, -distance and width) have been investi-gated. The advantages of the rear side emitter have been quantified with LBIC measurements (Fig. 1). The ge-ometry optimisation was assisted by 2D-computer simulations with DESSISTM. Different cell regions (busbar, fingers, etc.) were connected to a network in-cluding the series resistance of the contacts. A principle correlation between the EWT-design and observed low shunt values could be excluded by high resolution IR-thermography. The cell efficiency was increased to above 14 % by using a selective emitter, which can be realised within the EWT-concept without any further alignment steps. This is the highest efficiency world-wide reported so far for screen printed EWT-cells.
Fig. 1: LBIC measurement of an EWT-cell, arbitrary units, area 1 mm2. The small laser beam illuminates the front surface, the rear finger pattern is clearly visible due to the different collection probabilities of the rear side emitter (bright) with the connecting holes and the rear base area (dark).
A fruitful co-operation exists with BP Solarex and the Fraunhofer ISE in Freiburg within the ACE-designs project as well as with the MPI of Microstructure Physics, Halle.
Lamella-Silicon solar cells
Based on a macroscopically textured silicon substrate a novel high efficiency solar cell design for mono- and multicrystalline silicon photovoltaic devices has been developed, especially suitable for space and concentra-tion applicaconcentra-tions. It relies on cutting deep and narrow grooves with small distances between two neighbouring cuts. This results in a lamella-like appearance of the cell surface as seen in Fig. 2. The cutting is carried out with a conventional dicing saw equipped with a 15 µm thick saw blade.
After a shallow emitter diffusion and a thermal oxi-dation or silicon nitride deposition the dielectric is opened stripe-like at the top of one side of the lamellas by Shallow Angle Photolithography (SAP). The distance between the openings can be freely chosen by including lower lamellas between the ones with contacts. After a heavy phosphorous diffusion into the openings and the formation of a local back surface field the cells are met-allised by Shallow Angle Finger Evaporation (SAFE) 1) with or without SAP. This leads to very fine metal fingers.
Fig. 2: Acidic etched Lamella wafer.
The advantages of this novel solar cell concept are the very low reflection losses and the simple shallow angle photolithography with a mask-free light exposure.
Furthermore, the average distance between the place of optical generation of charge carriers and the collecting
emitter is extremely small like in vertical junction solar cells 2). This leads to very high collection probabilities of charge carriers even in materials with small minority carrier diffusion lengths.
Series of monocrystalline lamella cells and reference devices have been processed investigating the impact of different contact designs, especially by varying finger distances, screen-printed and photolithographically de-fined back surface fields (BSF) and shallow angle pho-tolithography conditions. Furthermore, the process se-quence for monocrystalline float zone silicon was adapted to the demands of multicrystalline silicon.
(1) B. Terheiden, P. Fath, G. Willeke and E. Bucher, Proc.
14th EPVSEC, Barcelona, 1997
(2) J. Lindmayer and C. Wrigley, 12th IEEE PVSC, Baton Rouge, Louisiana, 1976
Latest results on semitransparent POWER (Poly-crystalline Wafer Engineering Result) silicon solar cells 1)
Semitransparent crystalline silicon solar cells open new markets which were previously closed for photo-voltaics. They are especially suitable for applications in solar architecture (solar façades, stair cases, etc.) and for glass sliding roofs in the automobile industry. The optical transparency which results from a regular pattern of small holes in the mechanically engineered wafer can be varied in the range 0 - 30 % depending on the shape and size of the texturing tool (V-grooved or rectangular) as well as the distance between two neighboring grooves. In the case of bifacially active POWER silicon solar cells, the holes additionally lead to an electrical interconnection of the emitter on front and rear side.
The mechanical texturization technique 2) is inde-pendent of the structural properties of the starting mate-rial which may be cast or ribbon multicrystalline as well as single crystalline silicon. The grooves are created by using either a rectangular single blade (laboratory scale) or a texturization tool for industrial production, which allows the structuring of the whole wafer in only a few seconds 3).
Bifacial as well as monofacial semitransparent cell types were diffused with a 30 W/sqr. POCl3 emitter after the structuring of the wafers. A PECVD silicon nitride layer has been used as an ARC. The emitter on the rear side of the monofacial cells was removed after the SiN deposition. These cells do not have a surface passivation on the rear side and therefore show lower cell efficiencies compared to bifacial POWER cells. Finally, the screen printed metallization was fired through the antireflection coating.
Table 1 presents the results of illuminated IV measurements using the standard AM 1.5 spectrum. The 5 x 5 cm2 sized cells had a light transmittance of 18.2 % for the monofacial cells and 16 % for the bifacial cells.
Maximum efficiencies of 11.2 % for monofacial POWER cells and 12.9 % for bifacial POWER cells could be obtained on monocrystalline Cz-material with
different base resistivities. On multicrystalline silicon solar cells the open circuit voltage was found to decrease somewhat to values of around 560 mV, leading to efficiencies of 10.0 % and 11.1 % for the two cell types.
A little more elaborate processing including a silicon nitride with enhanced optical properties leads to the improved efficiencies in bifacial cells. Only six processing steps are necessary to produce monofacial POWER cells.
Tab. 1: IV-characteristics of different semitransparent POWER silicon solar cells. (cell area: 25 cm2)
.
Despite a semitransmittance of 18.2 %, a reduction of the collected current of only 8.1 % compared to conventional flat cells was observed on monofacially active multicrystalline devices. The good antireflection properties of the textured surface and the enhanced light-trapping of the cell geometry balance partly the optical losses due to the partial light transmittance. The open circuit voltage and the fill factor of both cell types are limited by a dominating second diode. The strongly enlarged length of the p-n-junction that enters the surface with a high recombination velocity increases the saturation current J02 of the second diode 4). Good sur-face passivation in these regions is therefore absolutely necessary to obtain high open circuit voltages as it is the case for the bifacially active cells.
The usage of base material with a reduced resistivity of 0.4 Wcm decreases the limiting saturation current of the second diode by a factor of almost two for Cz-mate-rial. Those monofacial cells showed an increase in VOC by 22 mV as compared to standard 1.0 Wcm material.
Due to the simpler processing 5,6), monofacial semi-transparent POWER cells seem to be more attractive for an industrial production which is undertaken by the company ‘sunways’ 7). Bifacial POWER cells, however, show a higher conversion efficiency. This can be explained by the good passivation of the rear surface of the solar cells. The emitter on the rear side covered with PECVD SiN leads to a reduced surface recombination and to a more efficient collection of minority charge carriers which are generated in the vicinity of the base contact. The spectral response in the long wavelength range is therefore strongly enhanced as compared to
monofacial semitransparent and conventional flat silicon solar cells.
(1) G. Willeke and P. Fath, 12th EPVSEC, Amsterdam (1994) p. 766
(2) G. Willeke, H. Nussbaumer, H. Bender and E. Bucher, Solar En. Mat. And Solar Cells 26 (1992) 345
(3) P. Fath, C. Marckmann, E. Bucher, G. Willeke, J. Slufcik, K.De Clerq, P. Duerinckx, L. Frisson, J. Nijs and R. Mertens, 13th EPVSEC, Nice (1995) p. 29
(4) R. Kühn, A. Boueke, M. Wibral, C. Zechner, P. Fath, G. Willeke and E. Bucher, 2nd PVSEC, Wien (1998) p.1390
(5) A. Boueke R. Kühn, M. Wibral, P. Fath, G. Willeke and E. Bucher, 2nd PVSEC, Wien (1998) p. 1709
(6) R. Kühn A. Boueke, M. Wibral, P. Fath, G. Willeke and E. Bucher, 2nd PVSEC, Wien (1998) p. 2415
(7) sunways AG, Macairestr.5, D-78467 Konstanz, Germany
hole hole
back metallization front metallization ARC ARC
emitter emitter
Fig. 3: Schematic drawings of monofacial (left) and bifacial (right) POWER cells.
Theoretical and experimental behavior of mono-lithically integrated crystalline silicon solar cells based on the HighVo solar cell concept
A new concept for the realization of monolithically integrated silicon solar cells has been presented 1-3). The concept is based on standard Si wafer technology and does not use thin film approaches. A key feature are isolation trenches dividing the wafer into several unit solar cells. Due to the imperfect isolation between unit cells UC defined on the same conductive wafer some new device aspects deviating from an ordinary se-ries connection of solar cells arise. For the theoretical description a model proposed by Valco et al. 4) has been generalized by using a two diode concept for the unit cells and by weakening the assumption of identical unit cells. The model was used to simulate the cell perform-ance in dependence on light intensity, isolation resis-tance, cell area and number of unit cells 2). As a result general design rules for these truly monolithically inte-grated solar cells are given. These design rules are quite general as they are also applicable on cells proposed in references 4 - 6. The theoretical predictions could be partially confirmed by experimental prototypes. The best cell with a total area of 21 cm2 and six unit cells exhibits an open circuit voltage of 3.5 Volts and a con-version efficiency of 12.8 % under 100 mW/cm2 AMG 1.5 illumination and standard reporting conditions.
The goal of any monolithic series connection of so-lar cells is to obtain a single photovoltaic device with output voltages not being limited by the energy band gap of the device's semiconductor material. Up to now the portable electronics market has been satisfied mainly by monolithic series interconnection of thin film solar cells although their conversion efficiency as well as their long time stability are still rather low compared to wafer-based crystalline silicon solar cells 7). For ap-plications with higher energy consumption several at-tempts to develop monolithically series connected solar cells without the help of a supporting substrate and thin silicon film technologies have been undertaken 4-6,8-10), but they all failed either physically or commercially due to the problems of cell isolation and/or cost intensive process techniques.
To achieve this goal we propose a rather simple ap-proach for defining, isolating and series connecting UCs which are very similar to standard low cost crys-talline silicon solar cells: Remove wafer material be-tween UCs by inserting trenches reaching from the cell front surface to the back surface. Remaining narrow bridges hold the device together. The trenches also de-liver a means for series connecting neighboring UCs by filling them with metal which additionally leads to an improved stability of the device. Fig. 4 and Fig. 5 illus-trate the cell design which we realized experimentally.
4 3
3 2
1 5
4
Fig. 4: Sketch of one Unit Cell UC and parts of the neighboring UCs. 1: silicon wafer, 2: emitter, 3: emitter contact metallization, 4: isolation trench, 5: base contact metallization.
3
1
4
5 2
Fig. 5: Cross section through the middle of the UCs from Fig 4. Series connection is achieved by printing the emitter metallization (3) through the isolation trench (4) to contact the base metallization (5) of the neighboring UC.
A key feature of the device structure is the remain-ing part of the wafer holdremain-ing the UCs together. Because of the wafer resistivities usually preferred in crystalline silicon photovoltaics (about 0.3 Wcm to 15 Wcm) these parts represent parasitic current paths severely influ-encing the device performance. Some of the conse-quences of these current paths can be predicted from an equivalent circuit model proposed by Valco et al. 4). Yet, for a better description of our cells we extended the one diode model in Valco’s equivalent circuit to the two diode model commonly used for characterizing Si solar cells (Fig. 6). Additionally we weakened the as-sumption of identical UC parameters by using a larger set of equations.
All UCs except the one which forms the negative pole of the device in forward bias exhibit some kind of additional shunting parallel resistance Rp across the wa-fer material. In our case Rp is given by the remaining
‘bridges’. The UC representing the negative pole (UC) is not shunted because the output current has to cross the pn junction of this cell to enter or leave the device through the emitter contact. Already from the model it-self two characteristic features of monolithically series connected solar cells are obvious. First, not only for Rp reaching infinity but also for M getting large the device behavior should be similar to that of an ordinary series connection; although in the latter case for UCs altered by Rp. The main influence of Rp is a reduction
Fig. 6: Equivalent circuit model for the description of monolithically series connected solar cells. The exter-nal wires represent the monolithic series connection which is obtained in our concept through the isolation trenches. Rp is given by the remaining ‘bridges’.
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 -200
-150 -100 -50 0
calculated from the equivalent circuit model
ideal series connection of the illuminated IV characteristics of one cell
with Rsh = 3000 9cm2 and five cells with Rsh = 67.31 9cm2
ideal series connection of six cells with Rsh = 67.31 9cm2
I [mA]
U monolithic array [V]
Fig. 7: Comparison of the equivalent circuit model for six UCs and low Rp with two approximations using an ordinary series connection
of the shunt resistance of the UCs leading to an effec-tive shunt resistance Rsheff. Rsheff dominates the device behavior at low illumination intensities. Second, the asymmetry in Rp in the device leads to an interconnec-tion of non identical solar cells which also effects the cell performance (Fig. 7). For Rp < 200 Wcm2 it could be shown that a small reduction of the area of UC1 with respect to the areas of the remaining UCs results in a gain in conversion efficiency which is in contrast to an ordinary series interconnection.
(1) S. Keller, P. Fath and G. Willeke, Patent application No. PCT/DE 99/00728 (March 1999)
(2) S. Keller, S. Scheibenstock, P. Fath, G. Willeke and E. Bucher, J. Appl. Phys. 87 (2000) 1556
(3) S. Keller, S. Scheibenstock, P. Fath, G. Willeke and E. Bucher, Technical Digest 11th International Photovoltaic Science and Engineering Conference, Sapporo City, (1999) p. 651
(4) G.J. Valco, V.J. Kapoor, J.C. Evans Jr., and A.T. Chai, Proceedings of the 15th IEEE Photovoltaic Specialists Conference, Orlando, (1981) p. 187
(5) V.J. Kapoor, G.J. Valco, G.G. Skebe, and J.C. Evans Jr., J. Appl. Phys. 57 (1985) 1343
(6) A. Goetzberger, US Patent No. 4 330 680 (May 1982).
(7) H.A. Aulich, Proceedings of the 13th European Photo-voltaic Solar Energy Conference, Nice (1995) p. 1441 (8) R.M. Swanson, US Patent No. 4 933 021 and 4 933 022
(Jun. 1990).
(9) M.A. Green, US Patent No. 4 323 719 (April 1982).
(10) U. Kerst and H.-G. Wagemann, Proceedings of the 14th European Photovoltaic Solar Energy Conference, Barcelona (1997) p. 2434
Back contact buried contact solar cells
Back contact solar cells offer various advantages over conventionally designed solar cells. Investigations by solar cell manufacturers showed that module assem-bling costs could be remarkably reduced due to the back contact design. Furthermore by placing the bus-bars on the back of the cell, the active front surface can be increased because of lower shadowing losses.
On the other hand the buried contact solar cell (BCSC) process 1) enables very low shadowing losses in conjunction with high quality front grid metallization and a selective emitter structure. This results in low se-ries and contact resistance and leads to a high short cir-cuit current density.
Combining the buried contact metallization and the back contact design we are presenting a new solar cell concept, the Metallization Wrap Through (MWT) con-cept. This concept is similar to the one presented by Kerschaver et al. 2) for screen-printed metallization.
The MWT solar cell has the photo carrier collection junction and the finger contact grid on the front surface which was defined by mechanical abrasion with 15 µm wide dicing blades. The MWT cell has one laser-drilled via in each n-contacting finger to transport the current to the n-type busbar on the back surface.
Co-diffusion of Al and P plays an important role in the processing sequence of our MWT cells. It allows to economise one furnace step by forming the Al-BSF and the heavy P-diffusion (POCl3 source) of the n-type grooves, vias, edges at the same time. The n-type bus-bars on the back of the wafers are protected by masks during the Al-evaporation.
The process sequence takes advantage of the selec-tive metal deposition during the electroless plating se-quence of the buried contact process. Since metal de-posits only on metal and semiconductor surfaces but not on dielectrics, the front SiNx ARC layer ensures that metal is only deposited within the grooves on the front surface. The n-busbars and p-contacts on the rear are fully metallized simultaneously. Thus the interconections i.e. the vias between the front grid and the n-contacts on the back are formed automatically during the plating step. The external p and n-contacts on the back of the cells are separated mechanically by thin sawing blades mounted on a dicing machine followed by an edge isolation step. Due to the combination of co-diffusion and mechanical contact separation no addi-tional p/n-contact definition steps are necessary in the MWT process sequence.
Fig. 8: Schematic cross-section of the MWT cell design.
The view is limited to the rear side n-busbar region
Fig. 9: REM picture of a laser drilled via after electro-less metal deposition in the whole via.
MWT cells (cell size 5 x 5 cm2
)
were processed at the University of Konstanz on CZ solar grade silicon which was part-processed (alkaline texture etch, POCl3 emitter diffusion, LPCVD-SiNx deposition) at the pro-duction line of BP Solarex in Madrid. An efficiencies of 17.3 % for an MWT (JSC = 37.3 mA/cm2, VOC = 612 mV, FF = 75.8 %, h = 17.3 %) shows that this back contact concept is quite promising. An in-crease of 3 % in JSC was obtained compared to the con-ventionally produced reference cells (JSC = 36.3 mA/cm2, VOC = 612 mV, FF = 76.2 %, h = 16.9 %). This gain can be explained by the shad-owing loss of the front busbar. VOCis almost identical for both cell types. The fill factors of the MWT cell shows that there was sufficient metal deposition in the vias indicating the potential of the buried contact proc-ess for back contact solar cells.(1) S.R. Wenham and M.A. Green: “Buried contact solar cells”, United States Patent No. 4.726.850, 1988 (2) E. Van Kerschaver, R. Einhaus, J. Szlufik, J. Nijs and