Options
Wafer sawing processes: from microscopic phenomena to macroscopic properties
Auteur(s)
Bidiville, Adrien
Editeur(s)
Ballif, Christophe
Rooij, Nico de
Date de parution
2010
Résumé
The majority of the solar modules are based on crystalline silicon (c-Si) wafers. Whereas the c-Si technology can offer a high efficiency, its main drawback (compared to traditional means to produce electricity) is the cost. The wafers account for one third of the total module cost, as high purity (thus expensive) silicon is required. One way to decrease the solar electricity cost is to decrease the amount of silicon needed for a wafer, for instance by sawing thinner wafers. Another way is to increase the solar cell production line yield by producing stronger wafers and hence reducing the breakage rate. From a technological point of view, both ways are equivalent, as thinner wafers have to be comparatively stronger to sustain the processing stresses. The core of this work is the analysis and understanding of the impact of wire-sawing parameters on the wafer’s mechanical properties with the aim of sawing thinner and stronger wafers. Wafer wire-sawing consists of a wire transporting a slurry, made of abrasive silicon carbide particles and lubricant, through a silicon brick. Such a process is complex, as it involves dynamic processes, fracture mechanics, fluid dynamics as well as tribological aspects on various length-scales. In the first part of the work, characterisation methods were developed to quantify the wafer quality: roughness, crack depth distribution, breakage stress, wafer thickness and Raman measurements were carried out. Furthermore, a TEM study was made to get a precise view of the silicon just below the wafer surface, and the surface at the top of the sawing groove was analysed to get a fundamental understanding of the material removal mechanisms. Two parametric studies were carried out to get insight into the influence of the sawing parameters on these wafer properties. The parameters that were studied were the abrasive size distribution, the slurry density, the wire tension and the feed rate. The first campaign focused on large parameter variations in order to have a global view of the variables determining wire-sawing, whereas the second campaign concentrated on lower variations and a more thorough study of these parameters closer to their standard values. From these data, mechanisms of crack creation are proposed and a novel semi-analytical model describing the wafer strength as a function of the sawing parameters is given. It is based on physical interactions and allows a deeper understanding of the sawing mechanisms. Furthermore, the effect of the sawing parameters on the wafer thickness is analysed. From these campaigns, it is seen that stronger wafers are obtained by using a fine abrasive, a low wire tension and a slow feed rate. A third study about the silicon debris impact on the wafer quality was carried out. It is found that below a given debris amount, they have no effect, but over this threshold, sawmarks appear and the wafer strength quickly decreases. This is put in relation with the findings from the two first sawing campaigns and a novel mechanism explaining the saw-mark creation is proposed: the debris prevents the abrasive particles from removing silicon as fast as required by the feed rate. This makes the wire pressure on the particle increase, until a second material removal mechanism appear. This mechanism is faster than the usual one and allow the wire bow (as well as the pressure on the particles) to decrease so that the sawing proceeds in fits and stops. From all the analysis done, a better general picture of the sawing process was gained. Differences between the top and the side of the sawing groove are explained, as well as differences between the wire entrance and exit side of the ingot. At the top of the groove, the roughness is lower and presents more facets and sharp angles than at the side (which is also the wafer surface). This is explained by a faster material removal rate at the top of the groove and by a lower maximal pressure on the particles. At the side, the particles can apply large pressure on the silicon when there is not enough room for the wire and several large particles overtaking each other. On the other hand, the slow material removal rate at the groove side leaves the smaller particles enough time to smooth the angles. The particles are progressively ejected from the groove side, which accounts for a roughness diminution and a wafer thickness increase in the first half of the wafer length. In the second half of the wafer length, the roughness is constant but the thickness still increases, indicating that the abrasive particle volume fraction in the slurry diminishes, without a notable change in particle size. Finally, near the wafer edges, the wire vibrations outside the silicon also account for a thickness deacrease. In the last part, diamond-wire wafering was studied as an alternative to the standard slurry sawing. The wafer surface characteristics were analysed and compared with slurrysawn wafers. On the diamond-wire sawn wafer, a thick layer of amorphous silicon was found. It has repercussions on topography, as smooth grooves are forming the surface, but also on the internal stress near the wafer surface: large stress has been measured by Raman spectroscopy and by EBSD, over 1 GPa. By applying the conclusions from this work, it is possible to saw thin and strong wafers. This study brings a better understanding of the material removal process at the micrometer level, helping the optimisation of the sawing. Furthermore, the developped semi-analytical model giving the impact of the sawing parameters on the wafer strength is a useful tool for producing stronger wafers. Finally, the presented study on diamond-wire sawn wafers brings relevant insights into the challenges that have to be faced before this technology can be successfully used by the industry.
Notes
Thèse de doctorat : Université de Neuchâtel, 2010
Identifiants
Type de publication
doctoral thesis
Dossier(s) à télécharger
