Engineering topics
Publications and source records attributed to Iles, P. A..
The basic principles for qualifying silicon wafering methods are summarized, and unconventional methods of wafering was discussed. Methods of cleaving analogous to diamond cutting, geological processes employing the expansion of freezing water, and karate chops are touched upon.
The orientation-dependent etching method of slicing was investigated to see if it could be used to form more slices from high quality silicon crystals than can be achieved by mechanical slicing methods. Orientation-dependent slicing uses preferential etching down narrow slots in a silicon slab on form slices. The method has several possible advantages including high slicing yield (sq m/kg), plane parallel, thin slices, ready for processing and the chance of high throughput and low costs. There are limitations in the need for simple crystals, and in restricted depth of slicing.
Fifty micron thick cells 2x4 sq cm area with coplanar back contacts were made with good yield, and with output equivalent to conventional top/bottom contact cells of the same thickness. A wraparound junction (WAJ) design was selected, and used successfully. The low alpha cells delivered were all above 12%, the average efficiency was 13% and the best was 14%. The overall yield was 35 to 40%, comparable to that for conventional 50 micron cells. The process sequence was moderately complex, but showed good reproducibility. The CBC cells performed wall under several important environmental tests. High alpha CBC cells were made, with about 1% increase in conversion efficiency. The most important design criteria were the choice of back surface N+ and P+ areas.
This paper describes cost effective production-ready space solar cells which can meet the requirements for use on the Space Shuttle and other large space missions. Actual yield and performance data for these cells, as well as cost comparisons between these and standard space cells are included.
Recent progress in thin (2-4 mils, 50-100 microns) silicon solar cells which extends the capability of the cells into areas of use for possible space missions is reviewed. The production of thin flexible arrays using wrapped-around junction and wrapped-around contact cells is considered. Attention is also given to the formation of cells with very low solar absorptance (alpha-s less than 0.66) through changes in the AR coating, changes in front contact area, and maximizing cell efficiency.
The state of the art technologies was applied to fabricate 50 micro thick 2x4 cm, coplanar back contact (CBC) solar cells with AMO efficiency above 12%. A requirement was that the cells have low solar absorptance. A wraparound junction (WAJ) with wraparound metallization was chosen. This WAJ approach avoided the need for very complex fixturing, especially during rotation of the cells for providing adequate contacts over dielectric edge layers. The contact adhesion to silicon was considered better than to an insulator. It is indicated that shunt resistance caused by poor WAJ diode quality, and series resistance from the WAJ contact, give good cell performance. The cells developed reached 14 percent AMO efficiency (at 25 C), with solar absorptance values of 0.73. Space/cell environmental tests were performed on these cells and the thin CSC cells performed well. The optimized design configuration and process sequence were used to make 50 deliverable CBC cells. These cells were all above 12 percent efficiency and had an average efficiency of -13 percent. Results of environmental tests (humidity-temperature, thermal shock, and contact adherence) are also given.
Work has progressed in fabrication and characterization of solar cells from ubiquitous crystallization process (UCP) wafers and LASS ribbons. Gettering tests applied to UCP wafers made little change on their performance compared with corresponding baseline data. Advanced processes such as shallow junction (SJ), back surface field (BSF), and multilayer antireflection (MLAR) were also applied. While BSF by Al paste had shunting problems, cells with SJ and BSF by evaporated Al, and MLAR did achieve 14.1% AMI on UCP silicon. The study of LASS material was very preliminary. Only a few cells with SJ, BSR, (no BSF) and MLAR were completed due to mechanical yield problems after lapping the material. Average efficiency was 10.7% AMI with 13.4% AMI for CZ controls. Relatively high minority carrier diffusion lengths were obtained. The lower than expected Jsc could be partially explained by low active area due to irregular sizes.
Solar cells were fabricated from EFG ribbons dendritic webs, cast ingots by heat exchanger method, and cast ingots by ubiquitous crystallization process. Baseline and other process variations were applied to fabricate solar cells. EFG ribbons grown in a carbon-containing gas atmosphere showed significant improvement in silicon quality. Baseline solar cells from dendritic webs of various runs indicated that the quality of the webs under investigation was not as good as the conventional CZ silicon, showing an average minority carrier diffusion length of about 60 um versus 120 um of CZ wafers. Detail evaluation of large cast ingots by HEM showed ingot reproducibility problems from run to run and uniformity problems of sheet quality within an ingot. Initial evaluation of the wafers prepared from the cast polycrystalline ingots by UCP suggested that the quality of the wafers from this process is considerably lower than the conventional CZ wafers. Overall performance was relatively uniform, except for a few cells which showed shunting problems caused by inclusions.
Two large cast ingots were evaluated. Solar cell performance versus substrate position within the ingots was obtained and the results are presented. Dendritic web samples were analyzed in terms of structural defects, and efforts were made to correlate the data with the performance of solar cells made from the webs.
To fabricate 50 microns thick, coplanar back contact (CBC) silicon solar cells, wraparound junction design was selected and proved to be effective. The process sequence used, the cell design, and the cell performance are described. CBC cells with low solar absorptance have shown AMO efficiencies to 13%, high cells up to 14%; further improvements are projected with predictable optimization.
Performance of solar cells made from various candidate silicon sheets is updated and the results are presented. Solar cells were fabricated using a baseline process and other process variations, and tested under AM1 conditions. Performance of the baseline solar cells indicates that remarkable improvements in material quality have been achieved for most of the sheets, showing efficiencies close to that of the control cells made from conventional CZ silicon. Process variations (or additions), in general, have resulted in significant improvements in all performance with a degree of change dependent on the process chosen and to a certain extent the sheet quality.
Solar cells from HEM, Dendritic Webs, and EFG ribbons were fabricated and characterized. The HEM solar cells showed only slight enhancement in cell performance after gettering steps (diffusion glass) were added. Dendritic webs from various growth runs indicated that performance of solar cells made from the webs was not as good as that of the conventional CZ cells. The EFG ribbons grown in CO ambient showed significant improvement in silicon quality.
Explore the source record for details and available documents.
Most effort was concentrated on development of procedures to provide large area (3 in. diameter) high efficiency (16.5 percent AM1, 28 C) P+NN+ solar cells. Intensive tests with 3 in. slices gave consistently lower efficiency (13.5 percent). The problems were identified as incomplete formation of and optimum back surface field (BSF), and interaction of the BSF process and the shallow P+ junction. The problem was shown not to be caused by reduced quality of silicon near the edges of the larger slices.
A copper based contact system using plated Pd-Cr-Cu was developed. Good cells were made but cells degraded under low temperature (300 C) heat treatments. The degradation was identified as copper migration into the cells junction region. A paper study was conducted to find a proper barrier to the copper migration problem. Nickel was identified as the best candidate barrier and this was verified in a heat treatment study using evaporated metal layers. An electroless nickel solution was substituted for the electroless chromium solution in the original process.
The plating sequences Pd-Cr-Cu and Pd-Ni-Cu are demonstrated. The surface was sensitized with a 50 A-thick Pd layer obtained from an immersion bath. After 15 min heating at 400 C in N2, a thin barrier layer of either Cr or Ni was deposited from electroless baths operated at temperatures around 90 C. The sintering process was repeated, and a thin copper layer of 500 A was deposited by electroless means. An electrolytic copper bath was used to build the copper layer to 3-4 micron thicknesses. Cells with good I-V curves were obtained, and the all-plated contacts had good adhesion. Preliminary cost estimates show that the process costs approximately 12 cents per watt excluding the cost of the masking procedure.
Explore the source record for details and available documents.
A technology assessment of present performance levels and current fabrication methods and designs is presented.