No full-text available
To read the full-text of this research,
you can request a copy directly from the author.
Wafering of Silicon
Abstract
Semiconductor bulk crystals have to be cut into wafers for further applications. The dominant slicing technique both for microelectronic and photovoltaic applications is the multiwire sawing method. The requirements on the processes and wafer qualities depend on the material and the application. The most advanced techniques have been developed for silicon. Sawing and the subsequent processes such as grinding, lapping, and polishing use abrasive particles for material removal. The fine-tuning and optimization of the wafer processes requires an understanding of the micromechanical interactions between abrasive particles and crystal. The current status of research and development will be described for the major methods and materials. Finally, a brief overview will be given for alternative wafer-processing techniques.
... The wire motion in this case is reciprocating (pilgrim-mode), and the wire cutting direction is reversed a dozen times during one multi-wire cut. A coolant fluid, such as water, is used in the DWS process [6]. ...
... In MWSS, the loose abrasives are not uniformly distributed inside the kerf over the contact length of the wire, making it more difficult to keep the surface integrity uniform within a wafer and to prevent the wafer from breaking during the cutting [7]. Compared to MWSS, the productivity achieved with DWS technology is two to three times higher, and it is possible to recycle the material removed from the kerf [6,8,9], as the Si debris is not mixed with SiC, unlike in MWSS. Although DWS shows important advantages over MWSS, there are some disadvantages such as corrugated surface shape formation, and diamond wire and silicon wafer breakage due to non-optimized cutting parameters [10]. ...
The wear of diamond-coated wire is an important cost driver in the multi-wire sawing of silicon-based photovoltaics. Understanding the different forms of diamond wire wear could help the solar energy industry to be more competitive by reducing wafer costs. In this study, an innovative method was applied to investigate the wear of diamond wires: several diamond grains in a diamond wire loop were tracked during the slicing of monocrystalline silicon. The grains attached to the wire were analyzed by scanning electron microscopy and focus variation microscopy. Based on the observed non-uniformity of wear on the perimeter of the cross-section of the wire, the rotation of the wire on its longitudinal axis was investigated. A new method that promotes wire rotation during cutting was successfully applied. The form and progression of the wear were also investigated by tracking single grains during the cutting process. Observations show the occurrence of nickel layer removal/deformation, micro-chipping of the diamond grains, abrasion wear, and sporadic grain pullout.
... With the development of science and technology, chip has become the basic energy for industrial production, and its quality is of vital importance. Silicon wafer processing is the basis of semiconductor manufacturing [1], [2]. Wafer defect detection is one of the key challenges facing the semiconductor manufacturing companies. ...
Silicon wafer is the raw material of semiconductor chip. It is important and challenging to research a fast and accurate method of identifying and classifying wafer structural defects. To this end, we present a novel detection method in terms of the convolution neural networks (CNN), which achieve more than 99% detection accuracy. Due to the wafer images are not available by open datasets, a set of imaging acquisition system is designed to capture wafer images. Digital image preprocessing technology is utilized to split a wafer image into thousands of silicon grain images. The proposed model, called WDD-Net, uses depthwise separable convolutions and global average pooling to reduce parameters and calculations, adopts multiple 1*1 standard convolutions to increase the network depth. Specifically, two types of CNN models, VGG-16 and MobileNet-v2, are adopted for comparative analysis. Using the aforementioned three models, the comparative experiments are implemented on data sets that consisting of more than ten thousand grain images. The experimental results show that compared with VGG-16 and MobileNet-v2, the detection speed of the WDD-Net is 105.6FPS, which is 5 times faster. The model size of the WDD-Net is 307KB, which is much smaller than the other two. Furthermore, the WDD-Net directly completes the data collection and defect detection process through the local computing equipment, which is suitable for edge computing.
This study aimed to evaluate and better understand the mechanical and crystalline responses of polycrystalline silicon sawn by diamond wire sawing. To simplify the multi-wire sawing kinematic, an endless wire saw with a single looped diamond wire welded was used. The wire cutting speed and feed rate were varied in order to evaluate the characteristics of surface morphology, surface roughness, and subsurface damage. The analysis of brittle-ductile transition and residual stress of the sawn surface and silicon chips were performed with Raman spectroscopy. The wear and failure mechanism of the diamond wire were analyzed. The results showed that sawn surface is composed of brittle and ductile regions and the predominance of one of these directly affected the surface roughness Ra. The ductile cutting mode induced the predominance of microgrooves and ploughing over the sawn surface and led to formation of an amorphous layer with residual compressive stress around of − 192.3 MPa. Micro-cracks in subsurface were identified and it reached a minimum depth of 7.2 ± 1.6 µm. Chip fragments and elongated chips were observed and latter is practically amorphous. The wire wear analysis indicated that during the cutting there is deformation of the Ni-layer, exposed grits, and grit pullout. The main wear mechanisms are Ni-matrix removal and abrasive wear of the diamond grit. A better surface quality of polycrystalline silicon was obtained on increasing wire cutting speed and decreasing feed rate. Thus, the looped diamond wire sawing allows to reach a high surface quality of the silicon wafer, since the material removal occurs more in ductile cutting mode, generating a smooth surface with shallow subsurface damage.
Graphical abstract
Diamond wire sawing (DWS) of silicon wafers has replaced loose abrasive sawing (LAS) within a very short time, mainly due to the enormous cost pressure in the photovoltaic industry. However, the LAS process is still much better investigated and understood from a mechanics point of view. This lack of micromechanically substantiated process knowledge is a major challenge in optimisation for a reliable process and the associated improvement in the quality of the products. The present work aims to contribute to this field by deriving a material removal coefficient that can be used at the process level in well-established material removal laws such as the Preston or Archard equation. The results show that the removal coefficient calculated on the basis of finite element simulations indicates a significant increase in material removal for common DWS sawing conditions in comparison to LAS, which is in good agreement with experimental findings.
To gain an in-depth understanding of the influence of diamond wire sawing of monocrystalline silicon on the feed force and wafer quality, experiments with a single diamond wire in the form of a loop were performed. The cutting parameters of wire cutting speed, feed rate and wire tension were varied. The feed force was monitored and surface roughness of the specimen was measured. The micro-crack depth was determined using an image processing. The results show that cutting conditions of higher feed rate and higher wire tension increased the feed force by 78% and 20%, respectively, since for chip removal these conditions is required a higher force per grain. On increasing the wire cutting speed, the feed force reduced by 66% due to a decrease in penetration depth of the grains. Based on an analytical model, it was evidenced that volume of material removed per wire length has a significant effect on feed force. The variation of the feed rate and wire cutting speed had a greater effect on surface roughness Ra and Rq, with a minor effect on Rz. With a higher feed rate the micro-crack depth increased, whereas it was reduced on increasing the wire cutting speed. A strong correlation between sawn surface and subsurface damage was found, indicating that the micro-crack depth is around 1.37 times higher than Rz value. The combination of a lower feed rate and lower wire tension with higher wire cutting speed resulted in lower feed force, smoother surface and shallower micro-crack depth.
Due to their excellent physical and mechanical properties, third-generation super hard semiconductor materials (such as SiC, GaN) are widely used in the field of microelectronics. From the crystal bar to electronic devices, slicing is the first machining procedure that directly affects the subsequent process. Fixed diamond wire saw has been widely used in cutting hard and brittle materials. However, the diamond grits of wire saw are bonded through the binding agent’s mechanical embedding that slicing super hard crystal is very difficult and inefficient. In order to improve the slicing efficiency, it is necessary to improve the holding strength and wear resistance of the diamond wire saw. The electro-spark deposition (ESD) process can form metallurgical bonding between metal materials at low heat input. The holding strength and wear resistance of the diamond wire saw can be effectively improved. In this paper, the mechanism of the manufacturing process of ESD diamond wire saw (ESDDWS) is introduced, and the conditions of the manufacturing process of ESDDWS are put forward. A model of the surface heat source of saw wire is established considering the wire shape. The transient thermal analysis of the single discharge of ESDDWS is carried out in ANSYS, and the effect of material compaction on material physical properties is considered. According to the simulation results, the parameter range of the manufacturing process of ESDDWS is predicted. The predictions agreed with experiment observation.
For decades, industrial companies have been collecting and storing high amounts of data with the aim of better controlling and managing their processes. However, this vast amount of information and hidden knowledge implicit in all of this data could be utilized more efficiently. With the help of data mining techniques unknown relationships can be systematically discovered. The production of semiconductors is a highly complex process, which entails several subprocesses that employ a diverse array of equipment. The size of the semiconductors signifies a high number of units can be produced, which require huge amounts of data in order to be able to control and improve the semiconductor manufacturing process. Therefore, in this paper a structured review is made through a sample of 137 papers of the published articles in the scientific community regarding data mining applications in semiconductor manufacturing. A detailed bibliometric analysis is also made. All data mining applications are classified in function of the application area. The results are then analyzed and conclusions are drawn.
The aim of this study was to investigate the influence of the cutting parameters on monocrystalline silicon cut by diamond wire sawing. The sawn surface was analyzed in terms of surface morphology, surface roughness, material removal mechanism and residual stress (by Raman spectroscopy). The surface morphology exhibited evidence of both material removal mechanisms: the brittle mode and the ductile mode. The surface roughness increased with a high vf, which promoted the formation of craters on the sawn surface. On applying a higher vc, the surface roughness reduced, since this favored the formation of damage-free grooves. The Raman spectrum showed evidence of different residual crystalline phases on the sawn surface, which confirms the material removal mechanisms. An increase in vf, for the same vc, caused at reduction in the compressive stress, since the brittle mode predominated as the material removal mechanism. Maintaining vf constant and increasing vc results in higher compressive stress, caused by plastic deformation of the silicon during chip formation.
At present, the diamond wire saw technology is applied to cut SiC single crystal. As a hard and brittle material, SiC slices may break in cutting due to local stress concentration and mutation. The void-defect, which is a typical defect inside the SiC single crystal can cause local stress concentration and stress field redistribution in wire sawing process. The degree of stress concentration is associated with the size and position of the defects, and understanding the sawing stress change of SiC single crystal with void defect during the wire sawing process is significant for the development of precision slicing technology. In this paper, a finite element model of wire sawing SiC single crystal containing spherical void defect was founded, and the stress concentration and sawing stress field distribution of SiC single crystal containing void defects of different relative positions and sizes have been analyzed. Numerical simulation results show that the stress concentration caused by the defects with different sizes at different relative positions in the cutting process is significantly different, and it does not always increase with the increase of defect size, but is relative to the ratio of defect size to kerf width or slice thickness. When the void defect is located inside the slice and infinitely close to the slice’s inner surface, this is the strongest influence of defect on slice stress concentration and stress field change during the sawing process, which means the possibility of slice breaking is large.
At present, diamond wire sawing technology is a mainstream method for wafering brittle materials such as silicon crystal, SiC crystal, sapphire, ceramics, optical glass and so on. Microcracks will occur on the wafers surface inevitably during sawing owing to the high brittleness and hardness of brittle material. The existence of microcracks and their size, orientation and distribution not only affect the surface quality and fracture strength of wafers, but also have an important impact on the subsequent processing effect of wafers after sawing. Therefore, the further investigation on the formation and propagation direction of microcracks is needed during diamond wire sawing. In this paper, an analytical model of the elastic stress field for brittle material is established during single abrasive scratching. The stress fields in front of and behind the indenter are analyzed respectively. Two different material removal modes, ductile removal and brittle removal, are considered. To evaluate the two different material removal modes, influence factor μ is proposed. The nucleation location and propagation direction of radial crack and median crack of several materials are predicted by analytical model. The result show that the radial crack nucleates behind the indenter. The deflection angle between propagation direction and scratching direction is about 35–60° and its size depend on the material removal modes. The nucleation location of median crack is no longer at the bottom of the plastic zone, but will deflect toward the scratching direction during scratching. The size of deflection angle depends on the tip half angle of indenter. When the tip half angle is 55–80° the deflection angle decreases with the increase of the indenter half angle. All predicted results are in good agreement with the experimental data of relevant scholars. At the end, the half-penny crack (radial/median crack system) presented in indentation process is extended to scratching process.
The dominating slurry based wafering technology for cutting multi crystalline silicon into photovoltaic wafers will be replaced by the diamond wire technology over the next years. Though, the slurry technology is still in use for manufacturing microelectronic and photovoltaic wafers, if wafer manufacturers are forced by different reasons to use their slurry based wire saws. The development of an enhanced slurry based process is still required. A Design of Experiments was performed varying two major wafering parameters, the wire diameter of structured wire as well as the silicon carbide particle size. Additionally, the table speed was subdivided in several process sequences to analyze its impact. As a result, the slurry particle size shows the major influence to the cutting efficiency and the wafer geometry. The F600 slurry shows the highest Preston coefficient, equal to a high efficient silicon removal, but also the highest standard deviation of wafer thickness. The F800 and F1000 show an equal silicon removal but the F800 causes the best geometry in combination with the investigated wire diameters. Nevertheless, different effects were observed that indicate a need for more detailed process analysis. The impact of the wire structure and its loss of structure due to the cutting process is not observed, yet.
A larger breakage ratio occurs with the decrease of wafer thickness due to the decrease of fracture strength for as sawn silicon wafers, which is a severe problem to limit the production yield of silicon wafers. It is necessary to understand the fracture behavior of as swan silicon wafers for increasing the wafer yield. In this paper, a predictive model for wafer fracture strength is proposed according to the linear-elastic fracture mechanics. The simulation results are comparable with the experimental results from references, indicating the correctness of this proposed model. Besides that, influences of surface crack parameters on the fracture strength of silicon wafer are discussed. Results indicate that the fracture strength of silicon wafer changes less when the surface crack inclination angle between the crack plane and saw mark is between 0 and 25°. However, the fracture strength increases with the increase of the surface crack inclination angle when the angle is larger than 25°. This proposed model can predict the strength distribution of silicon wafers, and it is very helpful to understand the fracture behavior of silicon wafers.
In order to optimize the process of wire sawing, this work studied the subsurface crack damage in silicon wafers induced by resin bonded diamond wire sawing using theoretical and experimental methods. A novel mathematical relationship between subsurface crack damage depth and processing parameters was established according to the indentation fracture mechanics. Sawing experiment using resin bonded diamond wire saw was performed on a wire saw machine. The validity of the proposed model was conducted by comparing with the experimental results. At last, the influences of processing parameters on subsurface damage depth were discussed. Results indicate that the median cracks are mainly oblique cracks which generate the subsurface crack damage. On the diamond wire saw cross section, the abrasives with the position angle 78° between abrasive position and vertical direction generate the largest subsurface damage depth. Furthermore, abrasives, generating the subsurface damage, tend away from the bottom of diamond wire with the increase of wire speed or decreases with the increase of feed rate. However, the wire speed and feed rate have opposite effects on the subsurface crack damage depth. In addition, the subsurface crack damage depth is unchanged when the ratio of feed rate and wire speed does not change.
Large attention is currently paid to the forthcoming diamond wire wafering technology, which shows a number of important advantages compared to the slurry based technology. Industrial diamond wire wafering solutions profit mainly from higher throughput, simpler processes and reduced production costs. However, the cutting mechanism affects the material properties, mainly the surface of the produced wafers and more importantly the wafer strength, which has an influence on the overall yield. The impact of the diamond wire technology on the wafer properties – in particular the wafer surface morphology – in comparison to the slurry based cutting technology is discussed in this article.
The usage of diamond-plated wire to produce silicon wafers for the photovoltaic industry is still a new and highly investigated wafering technology. The requirements regarding the quality of the wafer surface are very high and they have to compete with the cost effectiveness and quality of wafers produced by the established loose abrasive sawing technology. Hence, the wafer topography, the fracture stress and the corresponding sub-surface damage have to be investigated and improved. This paper discusses the topographic parameters, the crack depths and the fracture stress of mono- and multi-crystalline silicon wafers that were produced on multi-wire saws using diamond-plated wire and comparable process parameters. Especially multi-crystalline silicon (mc-Si) wafers exhibit lower fracture stress values compared to mono-crystalline silicon (cz-Si) wafers. We investigated the relations between crack depth and fracture stress. In detail, we determined a 15% higher median and a 40% increased interquartile range of the crack depth of mc-Si wafers in comparison to similar produced cz-Si wafers. That correlates with lower fracture stress values of textured mc-Si wafers compared to cz-Si wafers. In the following, we studied the sub-surface damage as a function of crystal orientation in detail. It was found that the crack depths increases from the {100} plane over the {111} plane to the {101} plane. However for the {101} plane two grains were investigated, resulting in a discrepancy of 4 μm. This may be related to the unknown rotation angle between the corresponding {111} cleavage planes and the wire direction and requires further investigations.
Wafers for the PV industry are mainly sawn with a multi-wire slurry saw. This process is slow (it takes almost half a day to complete a cut) and generates a lot of waste: around half the silicon is sawn away and contaminating the slurry, and the wire is worn and has lost strength. After each cut, the slurry has to be cleaned from the silicon debris and the wire has to be exchanged. In contrast, sawing the wafers with a diamond-plated wire is faster, requires only a cooling liquid that is easy to filter from silicon debris and uses a wire that can be kept for several cut. But this new sawing technique only has a chance to develop if the solar cell production lines developed for slurry sawn wafers is capable of processing these diamond-plated wire sawn wafers efficiently. This study focused on the differences of surface properties of wafers cut via a slurry wire-saw and via a diamond-plated wire-saw. From these surface differences, it is possible to explain the differences in cell processing behaviour and to update the cell production line. Finally, it is shown that wafers sawn with a diamond-plated wire can give cells that are as efficient as the slurry sawn wafers, which validates this new diamond-plated wire wafering method for the production of solar cells.
The growth of shaped silicon crystals has been attractive for a number of years and several methods to obtain various shapes, in particular flat ribbons, have been developed. Recently, however, the problem has been much more sharply focused due to the desire for the preparation of very low cost solar cells for terrestrial power generation, and a number of articles in the present volume address this problem.
IntroductionMultiwire Wafering TechniqueBasic Sawing MechanismExperimental ResultsSummaryReferences
Artificially grown crystals have to be cut into wafers for further applications. The current wafering techniques are described with a major focus on the dominant multi-wire sawing method. After sawing the wafers may have to be further treated by grinding, lapping, polishing, and etching procedures. The requirements on the processes and wafer qualities depend on the material and the application. The most advanced techniques have been developed for silicon, which is the major material for photovoltaic and microelectronic applications. Multi-wire sawing and the subsequent processes use abrasive particles for material removal. The fine-tuning and optimization of the wafer processes require an understanding of the micromechanical interactions between abrasive particles and crystal. The current status of research and development will be described for the major methods and materials. Finally, a brief overview will be given for alternative wafer processing techniques.
Abstract In the multi wire sawing process the wire tension is an important parameter for reliability and wafer quality. Changes of wire quality and wire tension from one end of a wire web in a multi wire saw to the other are investigated by several methods: wire tension measurements in the wire saw, wire elongation during an industrial process and material testing of representative wires. Two mechanisms for the loss of wire tension are identified. Firstly, the decrease of wire diameter due to abrasion causes a slow loss of wire tension along the whole wire web. Secondly, plastic deformation of the wire due to random overloading leads to a fast decrease of the wire tension within the first few centimeters of the wire web. Both effects are confirmed by modeling and calculations.
The effect of tool surface irregularities on the machining rate of the hydrodynamic polishing process is examined in this study. It is proposed that the tool surface irregularities will affect the film thickness or shear stress distribution between the tool and the work surface. Accordingly, the machining rate of this process is affected. Based on the elastohydrodynamic lubrication theorem and the machining mechanism of this process, the relationship between machining rate and tool surface topography is derived by the statistical analysis. It is shown that the machining rate has a linear relationship with the average variance of tool surface irregularities. The sensitivity of this relationship depends on the average film thickness. A large film thickness will result in a reduction of surface irregularities on machining rate. However, the randomness of machining rate due to the variation of tool surface topography cannot be effectively improved by increasing the film thickness. The experimental study shows that the qualitative and quantitative trends of machining rate verse tool surface irregularities can be well predicted by analytical analysis.
Free abrasive machining (FAM) is widely used for stock removal and surface finishing of ceramics. In FAM, material removal results from mechanical action between the abrasive slurry, which is trapped between the workpiece and a rotating lapping block, and the workpiece. Microscopic observations of the machined surface show that lateral cracking due to indentation by the abrasive particles contributes substantially to material removal. A simple model of FAM is developed which is based on indentation fracture and takes into account the abrasive particle distribution in the slurry. The model is used to predict the number of particles actually involved in the machining process, the distribution of load among these particles, and the depth of the plastically deformed layer on the workpiece surface. Many of the predictions of the model are well supported by experimental observations from the FAM of aluminum oxide, Ni-Zn ferrite, and glass using a silicon carbide slurry.
Ceramic materials are finished primarily by abrasive machining processes such as grinding, lapping, and polishing. In grinding, the abrasives typically are bonded in a grinding wheel and brought into contact with the ceramic surface at relatively high sliding speeds. In lapping and polishing, the ceramic is pressed against a polishing block with the abrasives suspended in between them in the form of a slurry. The material removal process, here, resembles three body wear. In all of these processes, the mechanical action of the abrasive can be thought of as the repeated application of relatively sharp sliding indenters to the ceramic surface. Under these conditions, a small number of mechanisms dominate the material removal process. These are brittle fracture due to crack systems oriented both parallel (lateral) and perpendicular (radial/median) to the free surface, ductile cutting with the formation of thin ribbon-like chips, and chemically assisted wear in the presence of a reactant that is enhanced by the mechanical action (tribochemical reaction). The relative role of each of these mechanisms in a particular finishing process can be related to the load applied to an abrasive particle, the sliding speed of the particle, and the presence of a chemical reactant. These wear mechanisms also cause damage to the near ceramic surface in the form of microcracking, residual stress, plastic deformation, and surface roughness which together determine the strength and performance of the finished component. A complete understanding of the wear mechanisms leading to material removal would allow for the design of efficient machining processes for producing ceramic surfaces of high quality.
Free abrasive machining (FAM) process associated with the wiresaw wafer slicing involves a three body abrasion environment. During the process, the cutting action is caused by fine abrasives freely dispersed irt the slurry, which get trapped between an axially moving taut wire and the ingot being sliced, bt this paper a model is proposed wherein the entry of abrasives into the cutting zone is governed by elasto-hydrodynamic (EHD) interaction between the slurry and the wile. An EHD film is formed by the abrasive carrying viscous slurry, squeezed between the wire and the ingot. This phenomenon is analyzed here using the finite element method. The analysis of such an interaction involves coupling of the basic Reynold's equation of hydrodynamics with the elasticity equation of wire. Newton-Raphson algorithm is used to formulate mid solve this basic coupling. The finite element discretization of the resulting nonlinear equation is carried out using Galerkin's method of weighted residuals. Basic hydrodynamic interaction model and the incorporation of the entry level impact pressure into the inlet boundary conditions are the two novel features introduced in this work. The analysis yields film thickness profile and pressure distribution as a function of wire speed slurry viscosity, and slicing conditions. A perusal of results suggests that the wiresawing occurs under "floating" machining condition. The minimum film thickness is greater than the average abrasive size. This is practically very important since the wiresaw is used to slice fragile semiconductor wafers with severe requirements on the surface finish. The possible mechanism by which a floating abrasive cart cause material removal is also touched upon in this work. Material removal rate has been modeled based on energy considerations. [S0742-4787(00)00702-5].
Wiresaw has emerged as a leading technology in wafer preparation for microelectronics fabrication, especially in slicing large silicon wafers (diameter greater than or equal to 300 mm) for both microelectronic and photovoltaic applications. Wiresaw has also been employed to slice other brittle materials such as alumina, quartz, glass, and ceramics. The manufacturing process of wiresaw is a free abrasive machining (FAM) process. Specifically, the wiresaw cuts brittle materials through the "rolling-indenting" and ''scratch-indenting" processes where the materials removal is resulting from mechanical interactions between the substrate of the workpiece and loss abrasives, which are trapped between workpiece and wire. Built upon results of previous investigation in modeling of wiresaw, a model of wiresaw slicing is developed based on indentation crack as well as the influence of wire carrying the abrasives. This model is used to predict the relationship between the rate of material removal and the mechanical properties of the workpiece together with the process parameters. The rolling, indenting, and scratching modes of materials removal are Considered with a simple stochastic approach. The model provides us with the basis for improving the efficiency of the wiresaw manufacturing process based on the process parameters.
Wire saw slicing is a cost effective technology with high surface quality for slicing large diameter silicon wafers. Though wire saws have been deployed to cut polycrystalline and single crystal silicon ingot since the early 1990s, very little is known about the fundamental cutting process. We investigate this manufacturing process and propose a contact stress model of wire saw slicing that illustrates the interactions among the wire, ingot, and abrasives (e.g., SiC) carried by the slurry. Stresses created by wire saw slicing silicon wafers are analyzed in this paper. During the cutting process, the wire moves at high speed (5-15 m/s) with respect to the silicon ingot. The abrasives in the slurry are lose third-body particles caught between the wire and ingot at the contact surface. The forces applied by the wire carry the abrasive particles and cause them to roll on the surface and at the same time to be constrained to indent the surface. Such rolling-indenting interactions result in the formation of isolated chips and surface cracks. The cracks and discontinuity on the surface also cause high stress concentration. As a result, the material is cut and removed. The stress fields of a single circular cone of the abrasive particle indenting on silicon crystal with normal and tangential forces can be calculated and analyzed from the modeling equations and boundary conditions. The stresses are expressed with dimensionless stress measures, as functions of normalized geometric parameters. The results show that the maximum normal stress occurs at the indentation point, while the maximum shear stress (sigma(zx)) occurs below the surface of contact, as expected. Such subsurface shear facilitates the peeling effects of the silicon cracks. Both the normal and tangential forces applied at the contacts are incorporated in the model. The model is very effective in explaining and predicting the behaviors and distributions of stresses during the cutting process, and can be used to determine the optimal geometry of the abrasive particles in the rolling-indenting process.
In wiresaw manufacturing processes, such as those in slicing silicon wafers for electronics fabrication, abrasive slurry is carried by high-speed wire (5 to 15 m/s), which exerts normal load to the surface via hydrodynamic effects and bow of taut wire. As a result, the abrasives carried by slurry are constrained to indent onto and roll over the surface of substrate. In this paper, the axisymmetric indentation problem in the free abrasive machining (FAM) is studied by modeling a rigid abrasive of different shapes pushing onto an elastic half space. Based on the harmonic property of dilatation, the closed-form solution of stress distribution inside the cutting material for three different indentation processes in common FAM process are presented: cylindrical and conical abrasives as well as uniform pressure distribution. Along the symmetrical axis, von-Mises stress is two times larger than that of local maximum shear stress for all three indentation conditions. The von-Mises stress is infinity at the contact point for sharp pointed indentation, a location of crack initiation and nucleation. For indentation by abrasive of flat surface, which also can be provided by the localized effects due to the hydrodynamic pressure acting on the surface, both the von-Mises and local maximum shear stress reach maximum underneath the contact zone.
A promising technique to form the silicon-on-nothing structure is presented as an alternative to the silicon-on-insulator structure. A large plate-shaped empty space in silicon (ESS) below the surface of the silicon substrate can be fabricated by connecting the spherical empty spaces, which are formed by surface migration of Si on the patterned Si substrate. The ESS technique has the potential to change the microprocess for the fabrication of large-scale integrated circuits and it can be applied to various manufacturing technologies.
Surface roughness measurements were performed on a glass-ceramic disk substrate by stylus profiler (SP), atomic force microscope (AFM) and non-contact optical profiler (NOP). Results of surface measurements are presented and the differences between SP, AFM and NOP roughness measurements are discussed. The effects of stylus size, scan size and sampling interval on roughness parameters are investigated. The methodology of choosing the scan size and sampling interval is suggested. AFM is concluded to be the most suitable surface measuring instrument for roughness measurement on the glass-ceramic substrate. If SP is used to make the measurement, the tip radius should be in the order of 0.2 mu m. However, localized damage to the test surface may occur owing to high contact stress. NOP using an objective magnification of 40 or lower is not recommended because the glass-ceramic substrate contains submicron roughness.
This paper presents a critical review and evaluation of knowledge of the grinding mechanisms for ceramic materials and their influence on the finished surface and mechanical properties. Two main research approaches are identified: a machining approach and an indentation fracture mechanics approach. The machining approach has typically involved measurement of the grinding forces and specific energy coupled with microscopic observations of the surface morphology and grinding detritus. Any proposed mechanisms of abrasive-workpiece interaction must be consistent with the magnitude of the specific energy and its dependence on the grinding conditions. The indentation fracture mechanics approach assumes that the damage produced by grinding can be modeled by the idealized flow system produced by a sharp indentor. Indentation of a ceramic body is considered to involve elastic/plastic deformation with two principal crack systems propagating from the indentation site; lateral cracks which lead to material removal and radial/median cracks which cause strength degradation. Each of these approaches provides insight into grinding behavior and strength degradation, but each has its shortcomings.
The recently reported hysteretic behavior of silicon under indentation (Clarke et al.1 and Pharret al.2-5) is investigated using an ultra-micro-indentation system with an 8.5 μm spherical-tipped indenter. The onset of “plastic” behavior during loading and hysteresis during unloading was readily observed at loads in excess of 70 mN. Cracking about the residual impression was observed only at loads of 350 mN and higher. An analysis of the data is presented that estimates the following: (1) the initial onset of deformation occurs at a mean pressure of 11.8 ± 0.6 GPa, (2) the mean pressure at higher loads is 11.3 ± 1.3 GPa, and (3) the hysteretic transition on unloading occurs at mean pressures between 7.5 and 9.1 GPa. These values are in good agreement with the accepted literature values for the known silicon transformation pressures. A simulation of the force-displacement data based on the analysis and model is presented and is found to fit the observations very well.
We present a both-sides-contacted thin-film crystalline silicon (c-Si) solar cell with a confirmed AM1.5 efficiency of 19.1% using the porous silicon layer transfer process. The aperture area of the cell is 3.98 cm2. This is the highest efficiency ever reported for transferred Si cells. The efficiency improvement over the prior state of the art (16.9%) is achieved by implementing recent developments for Si wafer cells such as surface passivation with aluminum oxide and laser ablation for contacting. The cell has a short-circuit current density of 37.8 mA cm−2, an open-circuit voltage of 650 mV, and a fill factor of 77.6%. Copyright © 2011 John Wiley & Sons, Ltd.
Multi-wire sawing is the main slicing technique for large multi- and monocrystalline silicon crystals in the photovoltaic and microelectronic industry. This paper describes the basic mechanisms by which slicing is achieved and develops a model for the material removal rate. It is shown that the hydrodynamic behavior of the slurry and the elastic interaction with the wire are an important aspect that has to be taken into account. The material removal occurs by the indentation of free floating SiC particles under the pressure of the wire. The microscopic fracture processes under the indented particles have been investigated and are described quantitatively. The numerical and experimental results are compared. (© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)
Silicon wafer wire-sawing experiments were realized with different sets of sawing parameters, and the thickness, roughness, and cracks depth of the wafers were measured. The results are discussed in relation to assumptions underlying the rolling–indenting model, which describes the process. It was also found that the silicon surface at the bottom of the sawing groove is different from the wafer surface, implying different sawing conditions in the two positions. Furthermore, the measured parameters were found to vary along the wire direction, between the entrance of the wire in the ingot and its exit. Based on these observations, some improvements for the wire-sawing model are discussed. Copyright © 2010 John Wiley & Sons, Ltd.
More than 80 % of the current solar cell production requires the cutting of large silicon crystals. While in the last years the cost of solar cell processing and module fabrication could be reduced considerably, the sawing costs remain high, about 30 % of the wafer production. At present the large crystals are cut using the multi-wire slicing technology[2] which has the advantage of a high throughput (several hundred wafers per day and machine), a small kerf loss of about 200 μm and almost no restrictions on the size of the ingots. Basic knowledge about the microscopic details of the sawing process is required in order to slice crystals in a controlled way. In the following the principles of the sawing process will be described in this review article as far as they are understood today.
The statistical fracture stress distribution of silicon wafer's was obtained by biaxial plate bending tests in combination with finite element calculations. For the correct interpretation of these tests it is important that the finite element calculations imply wafer thickness and elastic properties of the multicrystalline silicon wafer, otherwise the resulting stresses will be estimated to high. The Weibull distribution of fracture stresses yields different parameters for each test series of silicon, depending on the surface preparation and wafer manufacturing condition.
Nanoindentation has been used widely to study pressure-induced phase transformations in Si. Here, a new aspect of the behavior is examined by making nanoindentations on (1 0 0) single crystals using a series of triangular pyramidal indenters with centerline-to-face angles varying from 35.3° to 85.0°. Effects of indenter angle, maximum load, and loading/unloading rate are systematically characterized from nanoindentation load–displacement data in conjunction with micro-Raman imaging spectroscopy of the residual hardness impressions. Results are discussed in terms of prevailing ideas and models for indentation-induced phase transformations in silicon.
The “Epifree” process involves the lift-off of a high-quality monocrystalline film formed by reorganization upon annealing of cylindrical macropore arrays in silicon, and can thus provide high-quality silicon films without resorting to costly epitaxy. The challenge of this new process lies in etching controlled and regular pores in silicon in a cost-efficient way, and in developing a process compatible with the difficulty of handling a micron-thin material. Proof-of-concept cells have previously been achieved and this paper presents the latest progress, with a first development of thicker films and the inclusion of rear-side passivation. The energy-conversion efficiency of 1-µm-thin Epifree cells was improved from 2.6 to 4.1% by depositing a stack of amorphous silicon (a-Si) layers as rear-side passivation. The increase in Voc was, however, limited and bound to a drop in Jsc. The choice of a-Si was revealed to be unsuitable because of the thinness of the film and the presence of a full aluminum rear contact. The thinness of the film leads to a decrease in rear-side reflectivity by the a-Si absorption, and the aluminum, although not leading to crystallization, partly migrates inside the a-Si stack upon anodic bonding as shown by TEM. These factors indicate that an alternative surface passivation should be developed. In parallel to process developments, the material was thickened by modifying the macropore array dimensions, leading to a 2.4-µm-thick material over 1 cm × 1 cm areas. The efficiency of the next cells is expected to increase with this thicker material. Copyright © 2010 John Wiley & Sons, Ltd.
The application of indentation techniques to the evaluation of fracture toughness is examined critically, in two parts. In this first part, attention is focused on an approach which involves direct measurement of Vickers-produced radial cracks as a function of indentation load. A theoretical basis for the method is first established, in terms of elastic/plastic indentation fracture mechanics. It is thereby asserted that the key to the radial crack response lies in the residual component of the contact field. This residual term has important implications concerning the crack evolution, including the possibility of post indentation slow growth under environment-sensitive conditions. Fractographic observations of cracks in selected “reference” materials are used to determine the magnitude of this effect and to investigate other potential complications associated with departures from ideal indentation fracture behavior. The data from these observations provide a convenient calibration of the Indentation toughness equations for general application to other well-behaved ceramics. The technique is uniquely simple in procedure and economic in its use of material.
The phenomenon of apparent microhardness increase with increasing applied indentation test load, the reverse indentation size effect (RISE), was addressed from the viewpoint of indentation-induced cracking. The apparent microhardness when the cracking occurs was found to be related to the applied indentation test load as P
5/3. Previously published results on single crystals of silicon, GaAs, GaP and InP, which differ by a factor of four, all fall on the same line when analysed through this concept. It is concluded that the RISE is a result of the specimen cracking during the indentation.
The microhardness of Si (MP 1688 K), GaP (1623 K), GaAs (1510 K) and InP (1327 K) single crystals was determined by indentation (Vicker's hardness, VHN) of low-index facets at loads of 5–100g at 296–673 K, complementing earlier work on Ge and InSb. In the brittle range, extending up to about 0.35 T
melt (K), cracking occurred preferentially along the diagonals of the indentations, and was observed at all loads, with the possible exception of the lowest (5 g) in the case of InP at 289 K. At higher temperatures the relative orientations of crack and slip traces on the crystal surface, as observed by SEM, suggested that cracks nucleated preferentially at the slip-band intersection, as was also noted by Hirsch et al. (Phil. Mag.
3 (1985) 759) in GaAs above 600 K. As earlier in Ge, the VHN was found to depend on the load, L, as L
p
, and on the indentation diameter, d, as dn, with p = 1/2 and n = 2, as required by the model of indentation plasticity of Banerjee and Feltham [4, 5], but higher p and n values were found if chipping at the indentation edges was evident. The effect was related to the resulting decrease in indentation diameter due to the work lost, through chipping, by the indenter. Above about 0.35 T
melt (K), relaxation of the dislocation structures entails a decrease of p and n; both parameters tend to zero as T T
melt. Shear and tensile stresses seem to co-operate in the process of plastic deformation, the role of normal stresses, acting across slip planes, predominating in the brittle range.
Lapping of glass and other brittle materials is an important economic activity. Nevertheless, it has not received much scientific attention, despite the fact that it is also related to problems of wear (three-body abrasion). Therefore, lapping of glass has been analysed in terms of the concept of lateral fracture, by studying the influence of material parameters, namely Young's modulus, hardness and fracture toughness, on material removal and surface roughness for two different sets of experimental conditions. The concept was found to be well applicable, and was therefore used to develop a model of three-body abrasion by material removal via rolling and indenting abrasives. The model gives a good description of the two experiments. It allows an average normal force per abrasive to be determined from Preston's coefficient and the characteristics of the workpiece and abrasive.
A new implantation-free lift-off process is presented. We deposit a layer with mismatched thermal expansion coefficient with
respect to the substrate. Upon cooling, the differential contraction induces a large stress field which is released by the
initiation and the propagation of a crack parallel to the surface. The principle is demonstrated on both single and multi-crystalline
silicon. Films with an area of 25cm2 and a thickness of 30–50μm have been obtained. Some Si layers were further processed into solar cells. An energy conversion
efficiency of 9.9% was reached on a 1cm2 sample.
A concept that could provide a thin monocrystalline-silicon absorber layer without resorting to the expensive step of epitaxy would be very appealing for reducing the cost of solar cells. The empty-space-in-silicon technique by which thin films of silicon can be formed by reorganization of regular arrays of cylindrical voids at high temperature may be such a concept if the high quality of the thin film could be ensured on centimeter-large areas. While previous works mainly investigated the influence of the porous array on the final structure, this work focuses on the practical aspects of the high-temperature step and its application to large areas. An insight into the defects that may form is given and the origin of these defects is discussed, providing recommendations on how to avoid them. Surface roughening, pitting, formation of holes, and silicon pillars could be attributed to the nonuniform reactions between Si, SiO <sub>2</sub> , and SiO. Hydrogen atmospheres are therefore preferred for reorganization of macroporous arrays. Argon atmospheres, however, may provide high-quality silicon thin films as well, possibly even more easily transferable, as long as annealing is performed in controlled, clean, and oxygen-free conditions. Our experiments on large areas also highlight the importance of kinetics, which had not been considered up to now and which will require further understanding to ensure a complete reorganization over any wafer area.
In this paper it is shown that the viscosity of the liquid normal paraffins can be accurately defined as a simple function of relative free‐space except for values in the neighborhood of the freezing points of each compound. A novel method of extrapolating the specific volumes of this family of compounds to absolute zero is described which permits the calculation of reliable values of the relative free‐space from density data. An expression of the same form as the author's function, but in which temperature rather than free‐space is the primary variable (the so‐called Andrade equation), fails to reproduce the viscosity of n‐heptadecane over the same range of temperatures within the limits of the known accuracy of the measurements.
Silicon carbide abrasive grits were fractured dynamically in a roller crusher at ambient temperature and under controlled levels of atmospheric humidity, covering the full range from 10 to 100% relative humidity. The fracture stress depends strongly both on the grit size below a certain value and on the partial pressure of the water vapor. The size effect causes an increase in fracture stress as the grit diameter is decreased below a certain size. This is thought to be due to scaling of the size of pre-existing surface defects with grit diameter. The moisture effect causes a drop in fracture stress at any moisture level above 0% relative humidity (r.h.) for sufficiently small specimens, while the number of fragments created in the crushing process increases. This effect is thought to be due to moisture-assisted sharpening of the tips of surface defects, which serve as crack initiation sites, during the early stages of loading. The tip sharpening facilitates initiation of brittle fracture. It lowers the measured fracture stress and enables smaller defects to initiate secondary fractures when the stored elastic energy suddenly is released during the primary fracture; therefore, more fragments form. The results explain the previously observed moisture-assisted self-sharpening of abrasives which may result in increased abrasive wear rates for metals as the levels of atmospheric moisture increase. The results may also have implications for the interpretation of certain types of ceramic wear.
Lapping experiments on glass were performed to verify a model for three-body abrasion of brittle materials. The model is based on material removal by rolling abrasive particles. The particles indent the workpiece surface and remove material by lateral cracking. Median cracking introduces subsurface damage. The model leads to expressions for removal (or wear) rate, surface roughness, subsurface damage and load per particle as a function of particle shape, particle size, material parameters of workpiece and lapping plate, applied pressure and relative velocity between plate and workpiece. The model was found to give a good description of the experimental results, allowing among other things the calculation of removal rate, surface roughness or damage penetration from the measurement of either one of these parameters.
High-speed silicon ribbon growth offers the potential for substantial reduction of the manufacturing costs of photovoltaics. This paper outlines the new RGS process (RGS = ribbon growth on substrate) for the growth of silicon ribbons using a shaping die on a moving substrate. With pulling speeds in the range of 4–10 m/min, ribbons with a thickness of 250–350 μm and a width of 10 cm have been obtained. Columnar crystallization, grain sizes of the order of the ribbon thickness and up to millimeters, and favorable segragation effects offer the potential for solar cell efficiencies of up to about 10%.
The transfer of monocrystalline Si films enables the fabrication of efficient thin film solar cells on glass or plastic foils. Chemical vapor deposition serves to epitaxially deposit Si on quasi-monocrystalline Si films obtained from thermal crystallization of a double-layer porous Si film on a Si wafer. A separation layer that forms during this crystallization process allows one to separate the epitaxial layer on top of the quasi-monocrystalline film from the starting Si wafer after solar cell processing. Independently confirmed thin film solar cell efficiencies are 15.4% and 16.6% for thin film solar cells transferred to a glass superstrate with a total Si film thickness of 24.5 and 46.5 μm, respectively, and a cell area of 4 cm2. Device simulations indicate an efficiency potential above 20%.
Three-body abrasion is an important abrasive process in the surface finishing of ceramics and glasses. Generally it is assumed that material removal is achieved by rolling (indenting) particles. In this study, the removal rate of glass has been determined using SiC slurry as the abrasive and Cu as counter material. For large abrasive particle sizes the removal rate remains constant with time, while for small abrasive grain sizes it decreases. This is attributed to fracture and blunting of the abrasive particles. Displacement measurements revealed a decreasing ratio of bed thickness and mean abrasive particle size with particle size. The results are related to the particle size distributions of the abrasive powders used. On the basis of the particle size distributions, it can be concluded that the actual particle size distribution between workpiece and backing plate is different from the distribution of the slurry not lodged between both plates.
Chemical processes which occur during glass polishing are reviewed within the context of current mechanical models for the polishing process. The central chemical process which occurs is the interaction of both the glass surface and the polishing particle with water. A detailed mechanico-chemical model for the polishing process is proposed.