Wire for slicing a workpiece into slices by means of wire sawing

Abstract

The invention relates to a tensioned wire for simultaneously slicing a workpiece into a plurality of uniform slices by means of a wire saw and an abrasive slurry, wherein the wire has a plurality of crimps having two different undulation pitches, the amplitudes of which are identical at the selected tension

Description

[0001]Wire for Cutting a Workpiece into Slices by Wire Sawing: The invention relates to a wire for cutting a workpiece into slices by wire sawing. For a variety of applications, slices of uniform thickness and a high degree of plane parallelism of their front and back surfaces are required. Examples of such slices are those made of single-crystal silicon. A predominant method for producing particularly flat, uniform, and plane-parallel slices is wire sawing (multi-wire slicing, MWS). MWS cuts the rod into a multitude of uniform slices simultaneously in a single step, making it particularly economical. The present invention relates primarily, but not exclusively, to slices made of single-crystal silicon for use as substrate slices for microelectronic components.Such discs are also called "wafers" and are obtained by splitting a circular cylindrical rod of single-crystal silicon. Methods for wire sawing and suitable devices for this purpose are well known. In such methods, a wire is guided multiple times in a spiral around two to four cylindrical rollers (wire guide rollers) in such a way that a flat creel is formed between two of these rollers, consisting of parallel sections of wire (wire creel). The wire guide rollers have axles that are arranged parallel to each other and about which they are rotatably mounted. The outer surfaces of the wire guide rollers are provided with a multitude of ring-shaped, uniformly spaced grooves that run in planes perpendicular to the axes of the wire guide rollers and guide the wire.The workpiece, which is to be sliced ​​into discs, also has an axis aligned parallel to the axes of the wire guide rollers and is attached by adhesive to a device that feeds the workpiece onto the wire screen in a direction perpendicular to the plane of the wire screen. By rotating the wire guide rollers in the same direction around their axes, a relative movement is created between the wire segments and the workpiece. In the presence of an abrasive, each wire segment, upon contact with the workpiece and wire screen, removes material in the form of a kerf that forms in the workpiece. By continuously moving the wire screen, feeding the workpiece towards and through it in the presence of the abrasive, the wire screen slowly cuts through the workpiece perpendicular to its axis.On the side of the workpiece facing away from the wire frame, a sacrificial strip made of an easily machinable material is glued between the workpiece and the feed device. The slicing of the workpiece into wafers is complete when the wire frame is fully embedded in the sacrificial strip. The individual wafers are then connected to the partially cut sacrificial strip by the adhesive bond, like the teeth of a comb, so that the partially cut sacrificial strip and the fully sliced ​​workpiece can be easily removed from the saw as a single unit. By subsequently breaking the adhesive bond, the individual wafers are obtained, which can then be picked up for further processing and sorted into slots in a storage tray, the so-called wafer tray.Piano wire (hypereutectoid, pearlitic steel wire with a carbon content between 0.78% and 1.1%) with a round cross-section, a so-called smooth wire, is predominantly used. The shape of a smooth wire can be described as that of a straight cylinder with a circular base (wire cross-section) of a very large height (namely, the length of the wire). The wire has an axis (axis of the circular cylinder, longitudinal axis of the wire) and a diameter (namely, the diameter of the base). So-called structured or crimped wires can also be used for wire sawing. A structured wire is a smooth wire whose cross-sections have been displaced at numerous points along its wire axis by means of plastic deformation in directions perpendicular to the wire axis, so that numerous indentations and protrusions are created along the smooth wire axis.The indentations and bulges formed by plastic deformation of the smooth wire are also called crimps. The diameter of the right circular cylinder with the smallest base area that completely contains the structured wire is called the effective diameter of the structured wire, and the base area of ​​this cylinder is called the effective cross-section of the structured wire. The distance of the indentations and bulges from the longitudinal axis of the underlying smooth wire is then called the amplitude of the crimps, and the distance between two adjacent pairs of crimps is called the wavelength of the crimps, with each pair of crimps comprising exactly one indentation and one bulge. In this case, the effective diameter of the structured wire is the sum of the diameter of the starting smooth wire and twice the amplitude of the crimps. Wire sawing can be performed as so-called slurry wire sawing (SMWS) or diamond wire wire sawing (DMWS).In SMWS, abrasives, such as silicon carbide (SiC) or diamond, are suspended in a viscous carrier fluid, such as glycol or oil. This suspension is called a slurry. In DMWS, the abrasive, usually in the form of diamond, is firmly bonded to the wire surface, and a working fluid, usually water, possibly with a dissolved additive, such as a wetting and defoaming agent, is supplied to the wire. The working fluid itself contains no abrasives and acts as a cooling lubricant, counteracting overheating, detachment from the wire, and premature wear of the diamonds. It also helps to remove the chips from the workpiece produced by material removal from the parting line. SMWS is performed with smooth or textured wires, while DMWS is performed only with smooth wires.Structured wire forms "pockets" along its axis in the form of crimps, which act as reservoirs for the slurry fed to the wire during SMWS (Silicon Metal Sawing). This allows structured wire to transport more slurry into the cutting gap and evenly supply the cutting gaps of very large workpieces with correspondingly long maximum engagement lengths of the wire to the workpiece. The use of structured wire is also economical, as less slurry is required overall. This advantage generally outweighs the disadvantage of higher cutting losses due to the crimps. Structured wire is therefore of particular importance for cutting large silicon rods into substrate disks using slurry wire saws for particularly demanding microelectronic components. Structured wires and methods for their production are described, for example, in EP 1036235 B1, WO2012 / 069314 A1, and EP 3565693.From US 6,554,686 B1, it is known that it can be advantageous for a saw wire to be twisted, i.e., wound around its longitudinal axis. This can be achieved, for example, by first winding the wire onto an intermediate spool after the drawing process and then, using a winding arm rotating around the spool axis and a deflecting roller, removing it over the flange of the intermediate spool and winding it onto the final spool. By removing it over the flange, the wire acquires a 360° twist for each length corresponding to the circumference of one turn around the intermediate spool. For typical spool dimensions, such as those used for wire sawing and as described, for example, in US 8,418,951 BB, a wire twisted in this way exhibits approximately 1 to 2 twists per meter of wire length.The shape of a wafer can be described by the deviations of its front and back surfaces from the plane of regression of all points on the front and back surfaces, i.e., by two two-dimensional functions zVS(x,y) and zRS(x,y). VS stands for front and RS for back, y denotes the direction in which the rod is fed onto the wire frame during sawing, and x the direction in which the saw wire runs. Conveniently, the origin (x0=0, y0=0) is placed in the center of the wafer. The thickness of the wafer is then understood to be the function h(x,y) = zVS(x,y) – zRS(x,y), and the shape of the wafer is the function (x,y) = zVS(x,y) + zRS(x,y). TTV = max{x,y}(h(x,y)) then denotes, for example, the TTV (Total Thickness Variation, also GBIR, Global Backside Indicated Reading according to SEMI International Standards M65), where max(x,y) means the maximum over all points (x,y) within the disk.Due to the symmetries of the wire saw assembly, during wire sawing, zVS(x,y) and zRS(x,y) change far less with x for any given y than they change with y for any given x. Therefore, the one-dimensional functions h0(y) = zVS(x0,y) – zRS(x0,y) and s0(y) = zVS(x0,y) + zRS(x0,y) describe the thickness and shape profile of a wafer along the diagonal in the feed direction of the rod onto the wire saw almost as accurately as the two-dimensional functions h(x,y) and s(x,y), i.e., h(x,y)^h0(y), s(x,y)^s0(y). Thus, for a description of the wafer, it suffices to determine the height and shape profile of wire-sawn wafers by measuring h0(y) and s0(y). With ^^ = max. ^ Let ^^^ {^^(^ + ^) − ^^(^)} / ^ be the “waviness” of the shape gradient s0(y) with respect to a preselected window length L, where max ^ ^ ^^This means that the maximum within the interval y and y + L is to be determined. (For L ^^ 0, l ^i→m^^^(^, ^) = d^^(^) / d^ becomes the derivative of the shape profile w0(y) with respect to y.) Slurry wire sawing with structural wire known in the prior art generally leads to wafers that exhibit defects in their (linear) thickness h0(y), their (linear) shape s0(y), and their waviness w0(y), and are therefore unsuitable for particularly demanding applications. In particular, wires produced according to one of the methods specified in EP 1036235 B1, WO 2012 / 069314 A1, EP 3565693 B1 lead to wafers which, at certain cutting depths yRE, exhibit areas with greatly increased shape s0(yRE) and waviness wRE(yRE), whereby these defects occur spontaneously and unpredictably with otherwise unchanged cutting recipes (wire input, wire tension, wire speed,(Feed speed, etc.) only occur in some cuts and there at one or more unpredictable cutting depths yRE, and at these cutting depths can extend over a few adjacent wafers or to all wafers in the bar. The thickness h0(yRE) of the wafers at these defect locations remains largely unchanged compared to locations and wafers that do not exhibit these defects. That is, the width of the separation gap does not deviate significantly from that at defect-free cutting depths, and thus there is no different material removal from the front and back of the affected wafers, which form the walls of the separation gap. Due to these specific defect and occurrence characteristics, and the shape of h0(y) and w0(y) superimposed in a graphic at yRE, which visually resembles "rabbit ears",These specific geometric defects are referred to in the present invention as "rabbit ear" defects. Wafers with such "rabbit ear" defects are unsuitable for demanding applications, cannot be meaningfully processed further, and must be discarded as costly failures. The object of the present invention is therefore to provide a wire for simultaneously cutting a workpiece into a plurality of uniform slices by means of a wire saw, which produces slices that are free of "rabbit ear" geometric defects. This object is achieved by a wire for cutting a workpiece by means of a wire saw, comprising a straight wire with a circular cross-section and a wire longitudinal axis comprising a plurality of uniform pairs of plastic deformations in the form of displacements of the wire cross-section.wherein the first pairs of displacements have a first wavelength ^1 and a first amplitude A1 and lie in a first plane containing the wire axis perpendicular to the wire axis, the second pairs of displacements have a second wavelength ^2 and a second amplitude A2 and lie in a second plane containing the wire axis perpendicular to the wire axis, the first and second planes are perpendicular to each other and each pair comprises exactly one protrusion and one indentation, characterized in that the first and second wavelengths are different from each other, ^1 ^ ^2, and the first and second amplitudes are exactly the same at the selected tensile stress of the wire sections in the wire creel,A1 = A2. In one embodiment, the protrusions and indentations in each of the two planes have constant shapes and amplitudes, as well as constant distances from each other. Each pair can follow the path of a bending curve. The ratio of the sum of the wavelengths ^1 and ^2 to the magnitude of the difference between these wavelengths is preferably less than 30: (^1 + ^2) / |^1 – ^2| < 30. In another embodiment, the ratio of the least common multiple (“repeat length”) of the wavelengths ^1 and ^2, lcm(^1,^2), to the magnitude of the difference between these wavelengths is less than 500: lcm(^1,^2) / |^1 - ^2| < 300. In a particularly preferred embodiment, the two mutually perpendicular planes containing the wire axis, with the pairs of protrusions and indentations, are helically twisted together around the wire axis along the wire axis.wherein this twisting has a third wavelength ^3. The ratio of the least common multiple of ^1, ^2, and ^3 to the magnitude of the difference between the wavelengths ^1, ^3 is preferably less than 100,000 kgV(^1,^2,^) / |^1 - ^2|, where for ^3: 250 mm < ^3 < 2000 mm. The twisting can be carried out by elastic deformation. Likewise, the twisting can be carried out by plastic deformation. The wire preferably comprises a hypereutectoid pearlitic steel wire. The wire preferably has a carbon content between 0.85% and 0.95%. The diameter of the straight wire before the application of the deformations can have a diameter between 130 µm and 175 µm. The tensile strength is preferably between 3200 and 4500 MPa. To produce a wire according to the invention,A straight wire with a circular cross-section and a longitudinal axis is provided with a plurality of uniform pairs of plastic deformations in the form of displacements of the wire cross-section by passing it between the first teeth of a first pair of coplanar and meshing spur gears with parallel axes and subsequently passing it between the second teeth of a second pair of coplanar and meshing spur gears with parallel axes, wherein the axes of the first and second gear pairs are arranged perpendicular to each other, wherein the first pairs of displacements have a first wavelength ^1 and a first amplitude A1 and lie in a first plane containing the wire axis perpendicular to the wire axis, the second pairs of displacements have a second wavelength ^2 and a second amplitude A2 and lie in a second plane containing the wire axis perpendicular to the wire axis,The first and second planes are perpendicular to each other, and each pair comprises exactly one protrusion and one indentation, and the first and second wavelengths are different from each other, ^1 ^ ^2, and the first and second amplitudes are exactly the same at the selected tensile stress of the wire sections in the wire cutter, A1 = A2. Before applying the deformations, a multi-stage cold working process can be carried out by means of cold / wet drawing with a degree of deformation ^T between 3.40 and 3.75. The invention also relates to a wire saw comprising two or more cylindrical wire guide rollers rotatably mounted about axes, the outer surfaces of which are provided with a plurality of uniformly spaced, annularly closed grooves in planes perpendicular to the respective axis, around which the wire according to the invention is guided spirally in the grooves,that a flat gate of parallel wire sections is stretched between two wire guide rollers, facing the workpiece, with the wire sections held under a selected tensile tension, and a feed device to which the workpiece is attached and which can be fed perpendicularly onto the wire gate, the axes of the wire guide rollers and the workpiece being aligned parallel to each other. In one embodiment, the tensile tension of the wire sections in the wire gate is between 60% and 70% of the tensile strength of the wire. Such a wire saw using the wire according to the invention is suitable for simultaneously cutting a workpiece into a plurality of uniform slices by feeding the workpiece towards the wire gate and rotating the wire guide rollers in the same direction, so that the wire sections in the wire gate perform a relative movement to the workpiece.and the addition of a slurry of silicon carbide (SiC) or diamond in a carrier fluid of glycol or oil, such that when the wire creel and workpiece are brought into contact, material is removed which, with further wire movement, feed, and supply of the slurry, separates the workpiece into slices by means of a plurality of separation edges. The features specified with regard to the aforementioned embodiments of the wire according to the invention can be applied accordingly to the inventive method for wire production,The inventive wire saw and the inventive method for simultaneously cutting a workpiece into a plurality of uniform slices using a wire saw are described. These and other features of the embodiments of the invention are explained in the description of the figures and in the claims. The individual features can be implemented either separately or in combination as embodiments of the invention. Furthermore, they can describe advantageous embodiments that are independently patentable. List of reference numerals and abbreviations, 1 Draht2 Wire section 3 Left wire guide roller 4 Right wire guide roller 5 Axis of left wire guide roller 6 Axis of right wire guide roller 7 Rotation of left wire guide roller 8 Rotation of right wire guide roller 9 Wire feed 10 Wire discharge 11 Wire gate 12 Workpiece (silicon rod) 13 Movement of wire gate 14 Axis of workpiece 15 Putty strip (sacrificial strip) 16 Glue joint 17 Infeed 18 Rille 19 Left slurry nozzle strip 20 Right slurry nozzle strip 21 Slurry nozzle 22 Left slurry stream 23 Right slurry stream 24 Cutting depth 25 Separation notch 26 Marking notch 27 Amplitude A1 of the first crimps 28 Amplitude A2 of the second crimps 29 Wavelength ^1 of the first crimps 30 Wavelength ^2 of the second crimps 31 Intersection point 32 Thickness curve h0(y) (solid line) 33 Shape curve s0(y) (long dashed line) 34 Waviness w0(y) (short dashed line) 35 Double peak in waviness curve (rabbit ear defect) 36 Riefen37 Point of vanishing stretch / compression 38 Neutral fiber 39 Stretched area 40 Compressed area 41 Stretched edge fiber 42 Compressed edge fiber 43 Restoring force (stretch) 44 Restoring force (compression) 45 Crimp collapse 46 Isotropic wire wear 47 Half elliptical arc 48 Square root of sine function 49 Bending curve 50 Sine function 51 Area of ​​bending curve 52 Gear 53 Point load 54 Axis of gear 55 Crimps with large wavelength 56 Crimps with small wavelength 57 Point on the neutral fiber at crimp tip 58 Round effective cross-section 59 Oval effective cross-section 60 Axis of the underlying smooth wire 61 Point on the neutral fiber with vanishing curvature 62 Curvature profile of crimps made from half ellipses 63 Curvature profile of crimps with sin1 / 2x shape 64 Curvature profile of the Crimps with bend curve shape 65 Curvature profile of crimps with sinx shape 66 Cut wedge 67 Peak in shape curve (rabbit ear defect) 68 Increased wafer thickness (rabbit ear defect) 69 First wire wear 70 SecondWire wear 71 Constant wire wear 72 Preferential wire wear A1 Amplitude of first crimps A2 Amplitude of second crimps d, d' Diameter effective diameter Ð Spring constant in Newtons per millimeter, N / mm D Diffusion constant D0 Temperature-independent diffusion constant D(T) Temperature-dependent diffusion constant D.oC Depth of cut h(x,y) Thickness (two-dimensional) F Wire tensile force in Newtons, NGBIR Global Backside Indicated Reading (= TTV) h0(y) Thickness (one-dimensional, in feed direction) L Boxcar length k B Boltzmann constant (1.280649·10 -23 J / K ^1 Wavelength of the first crimps ^2 Wavelength of the second crimps ^3 Wavelength of the torsion ^ Lsmall wavelength of the beat from ^1 and ^2^H 1 2 ^ upper wavelength of the beat from ^ and ^ max^^^ Maximum function in the range l = 0 to l = Ls(x,y) shape (two-dimensional)s0(y) shape (one-dimensional, in feed direction)T thermodynamic temperature (in Kelvin, K)TTV Total Thickness Variation (= GBIR)w(x,y) waviness (two-dimensional)w0(y) waviness (one-dimensional, in feed direction)x wire directionx0(y) point on x (diagonal through disk center in feed direction)y feed directionzRS(x,y) distance of disk back from the regression planezVS(x,y) distance of disk front from the regression plane List of FiguresFig. 1 Functional elements of a wire sawFig. 2 Double crimped wireFig. 3 Wavelengths and amplitudes of a double-crimped wire Fig. 4 “Rabbit ear” defects and defect-free cut-off section Fig. 5 Defects of a wire with irregular crimping Fig. 6 Wear mechanism of a structured wire7 Wire crimp shapes and their curvatures Fig. 8 Cutting loss (kerf) vs. wire wear Fig. 9 Crimping device Detailed description of the invention First, the essential elements of a slurry wire saw (SMWS) are defined in Fig. 1 so that they can be referred to in the description of the invention. The SMWS process itself has already been described at the beginning. The workpiece 12 (silicon single crystal) with axis 14 has an orientation marking notch 26 milled into its circumference and is attached to a feed device (not shown) that allows feed in direction 17. A sacrificial strip 15 made of easily machinable material, for example graphite, is glued between the feed device and the workpiece (glue joint 16).Wire 1 is guided spirally along a direction 9 over at least two wire guide rollers 3 and 4 with axes 5 and 6 and grooves in grooves 18 under a longitudinal wire force (wire tensile force) such that sections 2 of the wire form a flat wire creel 11 facing the workpiece. The wire is taken from a fresh wire supply spool (payoff spool) and, after use, fed in direction 10 to a used wire supply spool (take-up spool) (both not shown). By rotating the wire guide rollers 7 and 8 in the same direction, a relative movement 13 is generated between the wire creel and the workpiece. The wire gate is supplied with slurry (suspension of abrasive grain in a viscous carrier fluid) via a left (19) and a right nozzle bar 20 with nozzles 21.By positioning the workpiece 12 (17) onto the moving (13) wire creel 11 and bringing the two into contact under slurry supply (22, 23), the wire creel causes material removal from the workpiece in the form of a multitude of separation notches 25, 24, where denotes the achieved cutting depth. Various structured wires, both with brass and zinc coatings, were manufactured by different wire manufacturers according to the inventors' specifications, and numerous test cuts were carried out on silicon rods with a diameter of 300 mm. The resulting wafers were measured with respect to their geometry, predominantly one-dimensionally (thickness h0(y), shape s0(y), waviness w0(y), and several other parameters) and also two-dimensionally on a sample basis (thickness h(x,y), shape s(x,y), waviness w(x,y), and several other parameters).The following describes the observations that led to the present invention and, based on these observations, describes the invention in detail. First, it was observed that structured wires, in addition to the material elasticity of their underlying smooth wire, exhibit elasticity due to their crimping; that is, an increase in wire tensile force is accompanied, in addition to their material elongation, by elongation due to the crimp structure in the direction of the wire's longitudinal axis. The inset in Fig. 3(a) shows the measured change in length of wavelengths ^1 and ^2 of two wires, each double-crimped in planes perpendicular to one another and perpendicular to the wire axis, at different wire tensile forces.The wavelengths 29 and 30 were, in both cases (a) and (b), during the tensile force-free measurement (0 N), ^1 = 2.65 mm (plotted on the left y-axis) and ^2 = 1.62 mm (right y-axis), and during the maximum tested tensile force (50 N), ^1 = 2.68 mm and ^2 = 1.64 mm. The most important finding was that the change in length of the longer crimps 29 and the shorter crimps 30 is different and thus exhibits two different spring constants ΐ with ΐ1 = 50 N / (2.68 – 2.65) mm = 1666.7 N / mm and ΐ2 = 50 N / (1.64 – 1.62) mm = 2500.0 N / mm. The difference between the wire in Fig. 3 (a) and that in Fig. 3 (b) was that the amplitudes 27 of the longer crimps and 28 of the shorter crimps in (a) and in (b) were chosen differently, such that the regression lines 27 (longer crimps) and 28 (shorter crimps) of the amplitudes A1 and A2 plotted against the tensile force in Fig. 3 (a) had an intersection point 31 at about 7 N and in Fig. 3 (b) at about 40 N, at which they are identical.Both wires were used in a wire saw cut on test bars with identical process parameters. In both cases, the wire was fed to the wire cutter with a wire tension of 34 N. With the wire from Fig. 3(a), wafers with a very irregular shape were obtained (comparative example), while with the wire from Fig. 3(b), wafers with a very flat shape were obtained (example according to the invention). During the cuts, the wire sections of the wire cutter exhibited a deflection in the feed direction of approximately 6 mm. Using the elastic modulus of steel (E = ~210 GPa; ^ = tensile stress [N / m2], ^ = elongation [m / m]) and the distance between the axes of the wire guide rollers of 550 mm (width of the wire gate, i.e. length of the bent wire sections) and an estimate of the frictional forces of the wire in the cutting gap when moving through the viscous slurry and by performing actual chip work, the actual tensile force of the wire sections in the cutting gap during the cutting process is calculated to be about 40 N.The wire in Fig. 3(a) therefore exhibited different amplitudes A1^A2 during the cutting process with an actual wire tensile force of 40 N, which varied depending on the tensile force. However, the wire in Fig. 3(b) showed approximately identical amplitudes A1^A2. Fig. 2 describes the properties of the structured wire in detail. Fig. 2(a) shows the wire on the left in a projection in the z-direction. The three spatial directions are defined here as follows: The x-direction corresponds to the longitudinal direction of the wire, i.e., the direction in which the wire moves during the cutting process; the y-direction corresponds to the direction of the bar feed onto the wire cutter; and the z-direction denotes the direction in which the workpiece axis runs. The wire had a diameter d of the underlying smooth wire of d = 175 µm and an effective diameter of deff = d + 2×A1, where A1 is the tensile force-dependent variable amplitude A1 from Fig. 3 (a) and (b), i.e., deff ^ 175 µm + 2×5 µm = 185 µm at 40 N wire tensile force according to Fig. 3 (b).The wavelength ^1 of the crimps 29 was approximately 2.675 mm (see inserts Fig. 3). Fig. 2 (b) shows on the left the wire in the y-direction projection with the second wavelength (30) ^2 and the second amplitude (28) A2. In Fig. 2 (a) and (b) the neutral fiber 38 of the wire as well as a point 61 with vanishing curvature (point on the neutral fiber 60 of the smooth wire underlying the structured wire before crimping) and a point 57 at the point of maximum crimping are also indicated, from whose vertical distance to each other the amplitudes A1 and A2 are read, as well as in Fig. 2 (a) the crimps with shorter wavelength from Fig. 2 (b) by 56, which in Fig. 2 (a) consequently run in a plane perpendicular to the plane of the drawing, and conversely in Fig. 2 (b) the crimps with longer wavelength from Fig. 2 (a) by 55, which in Fig. 2 (b) consequently run in a plane perpendicular to the plane of the drawing. On the right in Fig.2 (a) and (b) are each the projections of the wire in the x-direction, i.e. along the longitudinal axis of the wire. In the example of a wire according to the invention shown in Fig. 2 (a) and (b), where the amplitudes A1 and A2 are exactly the same at the selected wire tensile force, the effective cross-section of the structured wire is a circle 58, i.e., the effective cross-section of the wire is as "round" as possible, as is only possible with sequences of two crimps in mutually perpendicular crimp planes that are perpendicular to and contain the longitudinal axis of the wire. Fig. 2 (c) and (d) now show projections of two wires in the longitudinal direction (x) with an oval effective cross-section 59 of the wire, namely in Fig. 2 (c) a comparative example where the amplitude A1 of the crimps with the longer wavelength is greater than the amplitude A2 of the crimps with the shorter wavelength at the selected wire tensile force, and in Fig. 2 (d) a comparative example where A1 is smaller than A2.Undesirable oval cross-sections result, 59 which, during wire sawing, cause the wire to produce different widths of the cutting gap depending on its orientation, thus leading to undesirably irregularly shaped wafers. Secondly, it was observed that a saw wire twists around its longitudinal axis (self-torsion) when passing through the grooves of the wire guide rollers, i.e., the cutting gap in the workpiece. This could be observed directly and clearly, especially without a workpiece, by attaching a small piece of adhesive tape as a "flag" around a section of the wire and slowly running the wire over the wire guide rollers: The flag rotated randomly clockwise and counterclockwise.It was also observed that while the sum of all these left-hand twists at the end of the wire creel roughly corresponds to the sum of all random right-hand twists, it is usually not exactly identical, so that the wire is ultimately wound onto the old wire spool with a small remaining net twist. This twist causes a structured wire with two sequences of crimps, each in a separate plane, to appear "round" to the separation gap in the time-averaged mean of its movement along the wire's longitudinal axis (provided, as in the first observation, that the amplitudes of the two sequences of crimps are equal at the chosen wire tension). The wire can also be torsioned (intentionally / forced twisting around its longitudinal axis) during its manufacture to further promote this time-averaged round appearance.Such forced torsion can be achieved in two ways, or a combination of both. First, after drawing and crimping, the wire is wound onto an intermediate spool and then, using a deflecting device that rotates around the axis of the intermediate spool—the so-called flyer arm—is removed via the flange of the intermediate spool and wound onto a final spool. For each length of the circumference of a turn on the intermediate spool, the wire undergoes a 360° torsion. This typically amounts to between one and two torsions per meter. This type of torsion is called elastic torsion because, with such a small degree of wire twist, the wire is twisted elastically around its longitudinal axis, like a torsion spring, and no plastic deformation of the wire occurs. Second, during wire production, the entire crimping device can also be rotated around the longitudinal axis of the wire during the crimping process.In this case, crimps are directly imprinted onto the wire by means of crimping and the rotating crimping device; these crimps wind helically (spirally) around the wire axis. Since this is a plastic forming process, this type of torsion is also referred to as plastic torsion. Such torsion, already applied during wire production, forces an actual torsion during the cutting process in a preferred direction of rotation. Thirdly, it has been observed that wires with only one sequence of crimps in only one plane perpendicular to the wire's longitudinal axis, which contains the wire's longitudinal axis, can also result in wafers with good flatness.However, in this case, an additional torsion must be applied during wire production, and the wavelength^3 of this torsion (pitch of the helical rotation of the crimps around the wire's longitudinal axis) must be on the order of the wavelength^1 of the crimps,^3^^1, so that the wire appears sufficiently "round" in the slit over the time average of its movement along its longitudinal axis. This is practically very difficult to implement, however, since favorable crimp wavelengths are in the range of a few millimeters (see below), and an elastic torsion in the torsion process via an intermediate coil cannot be significantly higher than 2 / m (= 500 mm wavelength^3) or 4 / m (^3 = 250 mm) due to practical minimum dimensions of the intermediate coil. Therefore, elastic torsion would have to be applied several times in succession (perhaps up to ten times) to achieve the required short torsion wavelength.However, this repeated rewinding of the wire carries the risk of damage and is very uneconomical. With plastic twisting, at typical wire drawing speeds of 10 m / s or above, the rotation frequency of the crimping device around the wire axis would have to be more than 1 kHz (for example, for ³ = 10 mm), which is also technically impractical. Alternatively, the wire would have to be drawn very slowly, which is detrimental to wire quality (breakdown of the cooling lubrication). Both are either counterproductive to obtaining the high-quality wire required for good cut quality or very uneconomical. Therefore, only structural wires with crimps in two perpendicular planes (double crimping) were further investigated.Fourthly, it was observed that the wavelengths ^1 and ^2 of the crimps cannot be chosen arbitrarily, but must meet a number of criteria in order to obtain wafers with a particularly flat shape during wire sawing and, in particular, to avoid causing rabbit-ear defects. In a comparative example, a wire was used that had very closely spaced wavelengths ^1 and ^2, but ^1 ^ ^2. While this has the advantage that the amplitudes A1 and A2 change very similarly over a wide range of selected wire tensile forces, so that A1 always approximately equals A2 and the effective cross-section of the wire thus appears largely "round" in the average of the wire movement and the arbitrary or intentional torsion for the cutting gap, it led to numerous "rabbit-ear" defects that rendered the resulting wafers unusable.The inventors explain the occurrence of these rabbit-ear defects by suggesting that the (weak, ^3 >> ^1, ^2) torsions are "accumulated" by self-alignment of the wire in the cutting gap until the torsional restoring forces in the wire have increased to such an extent that they exceed the force of self-alignment in the cutting gap. At this point, the wire spontaneously and abruptly relaxes and its torsion dissipates. Since this occurs within any wire section of the wire creel and does not affect all wire sections simultaneously, the front side of a wafer, determined by the cutting gap where this wire relaxation takes place, experiences a different material removal rate at that point (at that cutting depth) than in the adjacent cutting gap, which determines the back side of that wafer.Wire relaxation due to accumulated torsion can extend across many adjacent slits; however, it is generally not identical for adjacent slits and thus for the front and back of each affected wafer. This explanation is further supported by the fact that "rabbit ear" defects at the affected cut depths always produce a slightly increased wafer thickness, meaning that less than average material is removed there, presumably because the rapid "re-twisting" renders the slurry reservoirs formed by the wire crimps ineffective. This results less in statistically random material removal by the suspended abrasive particles in the slurry and more in front- and back-differentiated damage to the wafer caused by friction from the largely dry, rotating wire.Different material removal rates and, in particular, surface damage to the front and back sides of a wafer lead to stresses in the wafer, which elastically deform the wafer at the affected location (the affected cutting depth). The phenomenon is limited to the locations (cutting depths) where spontaneous torsional relaxation occurs, and it can affect anywhere from a few to all wafers of the silicon rod – depending on how many cuts the torsional relaxation extends across. Such elastic deformations are converted into a permanent, non-elastic shape of the wafer in the machining step that follows the cutting, usually grinding or lapping, because the wafer, due to its elastic deformation, does not engage with the machining tool (grinding wheel, lapping plate) in a consistent area, resulting in uneven (preferential) material removal. Fig.Figures 4(a) and (b) show the measured shape (33) s0(y), waviness (34) w0(y) (both plotted on the left ordinate), and thickness (34)h0(y) (right ordinate) of two wafers of a section exhibiting "rabbit ear" defects 35 at different cutting depths (abscissa: depth of cut, DoC). Figure 4(a) shows the values ​​of a wafer from the rod end, i.e., near the fresh wire entry side of the rod, and (b) those of a wafer from the rod end (old wire exit side).4 (a) and (b) exhibit all the characteristics typical of a "rabbit ear" defect: - Primarily, the shape 33 and the waviness 34 of the wafer are affected at this location (peaks 67 in the shape curve, double peaks 35 in the waveness curve at the locations affected by the defect); - There is a slight increase in wafer thickness 32 at the affected locations 68, indicating reduced (abrasive) material removal and more likely surface damage; - Rabbit ear defects usually do not affect all wafers of a section, but often only a subset. In contrast, a defect in the wire itself, for example, uneven crimping, would always affect all wafers of the rod, since the defective wire section passes through all wafers of the rod sequentially. In the comparison example shown here, the wafers at the beginning of the rod (Fig. 4 (a)) are more affected than the wafers at the end of the rod (Fig. 4 (b)).The rod was very short at 202 mm physical length instead of the maximum possible length of 400 mm, so that with a longer rod, the wafers from the rod end would probably not have exhibited any rabbit ear defects at all. Figure 4(c) shows an example of a wafer from a section that is not affected by rabbit ear defects: Thickness 32, Shape 33, and Waviness 34 are comparatively uniform and without noticeable local deviations. By measuring the geometry of the wafers from test sections with different wires manufactured according to the inventor's specifications, it was found that wafers are free of rabbit ear defects when the least common multiples of their crimp wavelengths ^1 and ^2 are small compared to the wavelength ^3 of its (arbitrary or forced) torsion, lcm(^1, ^2) << ^3. The length of the least common multiple of ^1 and ^2 can also be referred to as the repeat length of the wire structuring. In the comparative example of the wire from Fig.In 4, the wavelengths were ^1 = 3.9 mm and ^2 = 4.1 mm, thus approximately ^1 ^ ^2, with lcm(^1, ^1) ^ 16 mm. In the series expansion of a periodic sequence of pairs, each consisting of a bulge and an indentation, the fundamental frequency of this period always dominates, since this sequence resembles a regular sine wave whose frequency is the reciprocal of this period and whose amplitude is equal to the amplitude of the bulges and indentations. A wire crimped in two planes with wavelengths ^1 (first plane) and ^2 (second plane) has the dominant frequencies f1 = 1 / ^1 and f2 = 1 / ^2, and thus resembles in form the curves of the functions sin(2^x / ^1) (first crimp plane) and sin(2^x / ^2) (second crimp plane), if the amplitudes are normalized to 1 for simplicity. In addition to the frequencies 1 / ^1 and 1 / ^2, the beat frequencies fL = |1 / ^1 – 1 / ^2| and fH = 1 / ^1 +1 / ^2 also occur.This follows immediately from the identity sin(2^x / ^1) + sin(2^x / ^2) = 2·sin(^x / ^1 + ^x / ^2)·cos(^x / ^1 - ^x / ^2). Accordingly, the beat wavelengths ^L = |^1 - ^2| and ^H = ^1 + ^2 also occur. It proved necessary for a wire free of "rabbit ear" defects that the beat frequencies not be too far apart. It proved advantageous if the quotient of the sum of the wavelengths ^1 and ^2 to the magnitude of the difference in the wavelengths of the two crimps is: (^1 + ^2) / |^1 - ^2| < 30. Furthermore, it proved advantageous if the repetition length of the wavelengths ^1 and ^2 (also normalized to the larger of the two beat frequencies, i.e., |^1 - ^2|) is limited. The repeat length is the shortest length at which the crimp pattern is exactly repeated by the (different) wavelengths ^1 and ^2, i.e., the least common multiple of the wavelengths ^1 and ^2, lcm(^1, ^2).For example, the least common multiple (LCM) of wavelengths ^1 = 3.8 mm and ^2 = 2.8 mm is 53.2 mm. In particular, the following always holds true: LCM(^1, ^2)^2·^2. Preferably, the wavelengths ^1 and ^2 are chosen such that LCM(^1, ^2) / |^1 - ^2| < 300. In the given example, 53.2 / 1 = 53.2 < 300. In a comparative example with closely spaced wavelengths ^1 = 3.9 mm and ^2 = 4.1 mm, (^1 + ^2) / |^1 + ^2| = 40^30, the least common multiple LCM(^1, ^2) = 159.9 mm, and LCM(^1, ^2) / |^1 - ^2| =799.5 ^ 300. Fifthly, it was observed that an optional forced torsion of the wire with a wavelength ^3 (length at which the two crimp planes are twisted exactly once around the wire axis, 360°) must be in a selected ratio to the wavelengths ^1 and ^2 of the crimps in the two crimp planes in order to produce, in particular, rabbit-ear defect-free wafers. ^3 must be sufficiently large compared to ^1 and ^2.It is preferred that ^1, ^2, and ^3 are chosen such that the ratio of their least common multiple to the magnitude of the difference between ^1 and ^2 is less than 100,000: lcm(^1,^2,^3) / |^1 - ^2| < 100,000. For ^3, wavelengths ^3 < 250 mm proved to be unstable to manufacture: With elastic torsion, ^3 < 250 mm would necessitate impractically small intermediate coils, and the flyer arm would have to rotate rapidly, or the rewinding process would take an uneconomically long time. With plastic torsion, the crimping unit would have to rotate so rapidly around the wire axis during the crimping process that any existing imbalances would lead to vibration and irregular crimping. On the other hand, wavelengths ^3 > 2000 mm proved ineffective compared to the inherent torsion of the wire. Therefore, it is preferable to choose ^3 such that 250 mm < ^3 < 2000 mm. Sixthly, it was observed that the wire properties (material composition, in particular...)The carbon content, the structure of the pearlite, and the degree of work hardening from the final (wet) drawing step must be as constant as possible along the entire length of the wire used in a cutting process. A test cut on a 360 mm long single-crystal Si(100) rod with a 300 mm diameter yielded the thickness 32, shape 33, and waviness 34 profiles shown in Fig. 5 for a wafer from the rod end (Fig. 3(a)) and for a wafer from the rod end (Fig. 3(b)) after a test cut. The structured wire used for the cut had wavelengths of ^1 = 3.8 mm and ^2 = 2.8 mm with a core wire diameter of 175 µm. The wire was fed to the wire gate with a tensile force of 34 N, which, including the observed wire deflection of 6 mm and the friction in the cutting gap, resulted in an actual wire tensile force of ~ 40 N during the cutting process.The crimp amplitudes were chosen to be equal at a wire tensile force of 40 N, A1 ^ A2 = 5 µm. According to the manufacturer, the wire should exhibit amplitudes over its entire length constant. The manufacturer only verifies this for test sections from the beginning and end of the wire winding of the wire spool used. The wire exhibited an elastic torsion of 1.25 / m on average (^3 = 0.8 m). Using a sensor consisting of two perpendicularly positioned optical micrometers (Keyence LS-9006), it was confirmed for several longer representative wire sections of the spool used for the test section that the core wire diameter and the crimp amplitudes within the test sections selected for verification were indeed very constant for the new wire over long lengths. Figure 5 now clearly shows high-frequency fluctuations in the wafer thickness of approximately 1 µm.These are also referred to as grooves 36 and are visually visible on the sawn wafer as a "washboard pattern" running in the feed direction. The sawing was performed using the pilgrim step method. The average wire length during one pilgrim step was 320 m. The cutting process took 13 hours, and the wire was moved longitudinally at a speed of 14 m / s. Thus, one pilgrim step (~2 × 320 m at 14 m / s) took ~45 s. Over the 300 mm rod feed length (rod diameter), > 1000 pairs of wire direction reversals (pilgrim cycles) were generated. The distance between two pairs of wire direction reversals was therefore < 0.3 mm on the surfaces of the resulting wafers. This can be determined by the measuring device to which the measurement diagrams in Fig.The ripples originate from the wafer and are not resolved by means of two opposing measuring probes, between which the wafer is passed and which capacitively determine the distance of one sensor to the wafer front and the other sensor to the wafer back, because the sensor diameter of Ø2 mm is significantly larger than the distances of the pilgrimage cycles on the wafer. The "ripples" 36 in the wafer thickness 32 in Fig. 5 (a) and (b) are therefore caused by the (worn) wire and not by the sawing process. After the separation process, the wire diameter of the worn wire used in the test cut was determined again with the laser micrometer. It was found that the diameter fluctuated between 163 mm (maxima) and 152 mm (minima) in time with the crimp wavelengths. In this thickness measurement, the crimps themselves are not visible; instead, only the diameter of the used wire (deformed in space by the crimps) is continuously determined.It can therefore be concluded that the structured wire – unlike a smooth wire – is subject to a dual wear mechanism: namely, isotropic (circumference-constant) wear of the underlying smooth wire of 175 µm – 163 µm = 12 µm and additional wear of the crimp tips of 163 µm – 152 µm = 11 µm. This second wear occurs anisotropically, namely only on the exposed sections of the crimps that come into direct material-removing contact with the workpiece. This is shown schematically in Fig. 6. Fig. 6(a) shows the wire 1 in its new condition. Due to the forming process of wire drawing, every (round) smooth wire exhibits internal wire stresses that increase radially from the neutral fiber to the outer fibers. These stresses are inherent to the drawing process and unavoidable. These stresses are radially symmetrical and consequently isotropic (angle-independent).Due to the additional deformation caused by crimping, a structural wire exhibits additional internal wire stresses. These are directly proportional to the wire's curvature resulting from the crimping process: The outer edge fiber 41 is stretched relative to the neutral fiber 38 in region 39, while the inner edge fiber 42 is compressed relative to the neutral fiber 38 in region 40, each by the amount resulting from the quotient of the local curvature and the distance of the respective fiber from the neutral fiber (half the wire diameter). This stretching 39 and compression 40 lead to internal tensile forces 43 and compressive forces 44, respectively, which—each individually—counteract the deformation but are precisely balanced, thus maintaining the stable shape of the crimped wire. At point 37, where the wire curvature disappears, exactly between two adjacent crimps, all additional stresses caused by the crimping process also disappear. Fig.Figure 6(a) also shows again the diameter d of the smooth wire underlying the structured wire (core wire diameter) and the effective diameter deff of the structured wire, which results from the sum of the diameter of the core wire and twice the amplitude of the crimps. It was observed that the plot of the kerf width against the cumulative length of the wire engagement for a wire not according to the invention does not have a constant slope, but exhibits a "kink": Figure 8(a) shows, as a comparative example, the 50% quantiles of the mean kerf widths (kerf loss) determined from 718 sections (mean sample size per section ~ 18 wafers, total > 13,000 wafers), plotted against the position of the wafer in the bar (0 mm = new wire entry side, 400 mm = old wire exit side of the maximum 400 mm long cut bars). For each millimeter step of the wafer position, the 50% quantiles were thus determined from the mean thickness of an average of 33 wafers.This averaging is necessary because otherwise the thickness-decrease curve would be so noisy that the aforementioned "kink" would not be visible. The kerf loss corresponds to the difference between the local wire creel pitch (center-to-center distances of adjacent wire segments in the saw wire creel). The kerf loss is calculated from the difference between the position-dependent creel pitch (known) and the mean wafer thickness measured at the respective creel position. This is identical to the sum of the effective diameter of the wire at the respective creel position and the mean thickness of the slurry film surrounding the wire in the cutting gap. The position in the creel corresponds to the cumulative length that the saw wire was engaged with the silicon rod at that position. The cutting gap width is therefore proportional to the effective diameter of the wire (if the thickness of the slurry film around the wire is approximately assumed to be independent of the wire's wear condition), and the wafer position in the rod (WP i. B.The wear rate of the saw wire is proportional to the wafer position in the batch. When the slope of this curve is determined, the (unknown) proportionality constants cancel out, and the slope directly indicates the wear rate of the wire with its cumulative engagement length. The wear rate is related to the effective diameter of the wire, i.e., core wire + 2 × amplitude of the crimps. The decrease in the width of the kerf loss shown in Fig. 8(a) exhibits a high first wear rate of ~0.094 µm / mm for the relatively fresh wire (i.e., for wafer positions near the fresh wire entry side of the rod) and a second, lower wear rate of ~0.039 µm / mm for the relatively old wire (i.e., for wafer positions near the old wire exit side of the rod). The relatively new wire in a comparison example thus wears out approximately 2.4 times faster than the relatively old (worn) wire. Fig.Figure 8(b) shows, as an example, the progression of the separation gap width (~ wire wear) against the wafer position in the rod (~ cumulative wire engagement length). This wire exhibits a constant wear rate 71 from the fresh wire entry to the old wire exit side of ~ 0.77 µm / mm. Figure 6(b) explains the undesirable wear behavior of a wire not according to the invention. If the wire has a high curvature in the crimp region, i.e., high restoring forces 43 of stretching in the stretched region 39 and high restoring forces 44 of compression in the compressed region 40, then the equilibrium of these forces is disturbed by preferential (one-sided) material removal from the stretched region, and the restoring force 44 of compression 40 predominates (Figure 6(b)). As a result, the structure of the wire partially collapses (45).A non-inventive structural wire with high wire curvatures in the area of ​​the exposed crimps is thus subject, in addition to the anisotropic wear 72 of the crimp tips and the isotropic wear 46 of the core wire, to an additional reduction 45 of its effective diameter. This partial collapse of the crimps reduces the amplitude of the crimps in addition to their double wear (isotropic and anisotropic), and the structural wire loses its effectiveness to an even greater extent in transporting sufficient slurry into the separation gap by means of the slurry reservoirs formed by its crimps. Such a collapsed structural wire behaves more like a smooth wire, and wire saw cuts performed with such a wire produce wafers with uneven material removal and thus uneven thickness profile 32 (compare Fig. 5) in the form of "saw marks" 36 (grooves).A preferred option is therefore a structured wire that exhibits as little internal stretching and compression stress as possible in the area of ​​the crimp tips, i.e., low curvature. Fig. 7 (a) shows various shapes ^(^) for each pair of adjacent crimps, i.e., exactly one wavelength ^, namely (curve 47): in the form of half ellipses joined ^to one another, ^el(^) = ^1 − (2^. ^ ^ − 1) ^ ^^ for 0 ^ x ^ ^ and^ ^el(^) = −^1 − 1) ^ ^^ for ^ < x ^ 2^ (solid curve);- (curve 48): a sin1 / 2x-shaped curve,^(^) = +√sin ^ for 0 ≤ ^ ≤ ^, ^(^) = −√sin ^ for ^ ≤ ^ ≤ 2^ (long dashed curve);- (curve 49): a curve following a deflection curve, (short dashed curve), where the modulus of elasticity E, area moment of inertia I, and bending force F were normalized such that the amplitude assumes the value 1 and the wavelength the value 2^ so that the curve shapes are directly comparable;- (curve 50): and finally a sinusoidal curve 50,^(^) = sin ^ (dotted curve). Fig. 7 (b) shows the curvatures of the wire corresponding to the shape profiles f(x) of the crimps in 7 (a), specifically for the profile composed of semi-ellipses (solid curve), 62, the sin 1 / 2 x-shaped profile (long dashed curve), 63, the profile according to bending curves (short dashed curve), 64, and finally the curvature profile 65 for the sinusoidal crimp profile (dotted curve). The curvature ^(x) is, according to its definition, from ^(^) = ^ ^ ^ calculated, where f(x) is the course of the curve, ^^(^) = ^^^ + ^^(^) is an infinitesimal element of the arc length ^(^) = ∫ ∙ ^^ denoted and ^^(^) = the first and The second derivative of the curve f(x) with respect to position (x-axis). Alternatively, the curvature ^(x) can also be determined numerically from the actually measured curve for the crimps. It turns out that the curve 47 (solid curve) composed of half-ellipses for the crimps at the exposed crimp tips at and3^^ 2 and, alternatively (very similar curve and curvature profile), the sin 1 / 2The x-curve 48 exhibits the smallest curvatures in magnitude 62 and 63, but very high curvatures with abrupt transitions at the zero crossings at ^ = 0, ^ = ^ and ^ = 2^. Seventhly, it was observed that the structure imparted to a smooth wire by asymmetric plastic deformation (crimping) is not constant in its amplitude over time: The amplitude of the crimps measured immediately after the crimping process does not correspond to the amplitude measured after a storage period. The amplitudes of the crimps change slowly with the storage time. In doing so, they can either "grow" (the effective diameter of the wire increases over time) or "shrink" (the effective diameter of the wire decreases over time).This was determined by examining sections of different wires from the same production batch, but with varying wire specifications, coatings (metal or zinc), and from different manufacturers. These sections were cut shortly after wire production and again after extended storage periods. In cases where the wire's effective diameter increased over time, the mean wafer thickness decreased, and conversely, in cases where the wire's effective diameter decreased over time, the mean wafer thickness increased. The growth or shrinkage rate is initially highest and then decreases over time, so that after extended storage (approximately 6 months at room temperature), the wire reaches a stable final effective diameter. This behavior often goes unnoticed because shipping and storage time account for a significant portion of this period before the wire is actually used.Whether the effective diameter of a wire "grows" or "shrinks" over time appears to depend on the concentration of internal wire stresses. The inventors explain this behavior with diffusion processes: Wire stresses exist in the form of lattice defects (dislocations and point defects such as additional or missing atoms). These represent an imbalance compared to adjacent areas in the wire with lower stresses, the gradient of which drives diffusion. Even below the activation energy for diffusion (a few hundred °C), since the diffusion coefficient D(T) follows the Arrhenius equation, Δ(Δ) = exp(ΔA / ΔB), the internal stresses can diffuse even at room temperature – albeit at a significantly slower rate than at higher temperatures – thus leading to a slow, time-dependent deformation of the wire.Here, D(T) denotes the temperature-dependent diffusion constant, T the thermodynamic temperature, and D0 the temperature-independent diffusion constant (unit m). 2 / s), EA the activation energy and k B The Boltzmann constant. A crimp shape with elliptical or sinuous 1 / 2X-shaped profiles, which exhibit low curvatures and thus low internal stresses due to plastic deformation at the exposed crimp tips, but high and abruptly changing curvatures and thus stresses at the zero crossings (extrema with sign changes at the zero crossing), apparently lead to a similar partial collapse of the crimp structure as shown in Fig. 6(b), only due to deformation at the zero crossings of their structure. The time-dependent shape change of the wire occurs because the stretching and compression forces, whose restoring forces balance each other and initially stabilize the shape of the wire, exhibit different concentration gradients, are thus subject to different diffusion, and are therefore degraded unevenly, causing the wire to change its shape.The process comes to a standstill and the wire reaches its long-term stable shape when the stress concentrations have decreased sufficiently through diffusion that the remaining concentration gradients are no longer adequate to drive further diffusion. Therefore, with reference to Fig. 7, crimp shapes f(x) that follow the curve of a bending pattern are preferred. Furthermore, a bending curve is the easiest to generate. This is shown in Fig. 9: The distances between the axes 54 of the two gears 52, which deform the smooth wire into a structured wire, are chosen such that the wire, when passing through the gears, is supported on three points 53, namely a pair of gear tips of one gear and a gear tip of the opposite pair, and is thus deformed according to a bending curve, without, however, being pressed so strongly into the tooth valleys that the entire tooth shape is imprinted on it. In the latter case, the required deformation forces would be so large that the cross-section of the wire would be altered – the wire cross-section would be partially flattened and thus become non-circular, which would be detrimental to uniform material removal in the cutting process and to its forming behavior in the second crimping step. With this deformation according to a bending curve, the shape of the tooth flanks and tooth valleys plays no role.The tooth tips only need to be rounded enough so that the wire is not notched at the contact points during the forming process. In particular, a perfectly sinusoidal deformation of the wire, which is usually given in the prior art as an example shape for a structured wire, cannot be produced without changing the wire cross-section. This is undesirable because the second crimp, which takes place in a plane perpendicular to the plane of the drawing in Fig. 9, would then form a non-round wire, resulting in a different profile shape for the second crimps than for the first crimps. This would lead to asymmetrical wire wear during sawing and to wafers with uneven thickness and shape. The above description of exemplary embodiments is to be understood as illustrative.The disclosure thus rendered enables the person skilled in the art, on the one hand, to understand the present invention and its associated advantages, and, on the other hand, also encompasses, in the understanding of the person skilled in the art, obvious modifications and alterations of the described structures and methods. Therefore, all such modifications and alterations, as well as equivalents, are intended to be covered by the scope of protection of the claims.

Claims

1. Claims 1.Wire for cutting a workpiece by means of a wire saw, comprising a straight wire with a circular cross-section and a wire longitudinal axis, comprising a plurality of uniform pairs of plastic deformations in the form of displacements of the wire cross-section, wherein the first pairs of displacements have a first wavelength ^1 and a first amplitude A1 and lie in a first plane containing the wire axis perpendicular to the wire axis, the second pairs of displacements have a second wavelength ^2 and a second amplitude A2 and lie in a second plane containing the wire axis perpendicular to the wire axis, the first and second planes are perpendicular to each other and each pair comprises exactly one protrusion and one indentation, characterized in that the first and second wavelengths are different from each other, ^1 ^ ^2, and the first and second amplitudes are exactly the same at the selected tensile stress of the wire sections in the wire cutter, A1 = A2.2.Wire according to claim 1, wherein the protrusions and indentations in each of the two planes have constant shapes and amplitudes as well as constant distances from each other.

3. Wire according to claim 1 or 2, wherein each pair follows the path of a bending curve.

4. Wire according to any of the preceding claims, wherein the ratio of the sum of the wavelengths ^1 and ^2 to the magnitude of the difference between these wavelengths is less than 30: (^1 + ^2) / |^1 – ^2| < 30.

5. Wire according to any of the preceding claims, wherein the ratio of the least common multiple (“repeat length”) of the wavelengths ^1 and ^2, lcm(^1,^2), to the magnitude of the difference between these wavelengths is less than 500: lcm(^1,^2) / |^1 - ^2| < 300. 6.Wire according to one of the preceding claims, wherein the two mutually perpendicular planes containing the wire axis, with the pairs of protrusions and indentations along the wire axis, are helically twisted together around the wire axis, and this twisting has a third wavelength ^3.

7. Wire according to claim 6, wherein the ratio of the least common multiple of ^1, ^2, and ^3 to the magnitude of the difference between the wavelengths ^1, ^3 is selected to be less than 100,000, lcm(^1,^2,^) / |^1 - ^2|, and simultaneously 250 mm < ^3 < 2000 mm.

8. Wire according to one of claims 6 or 7, wherein the twisting is carried out by means of elastic deformation.

9. Wire according to one of claims 6 or 7, wherein the twisting is carried out by means of plastic deformation.

10. Wire according to any one of the preceding claims, wherein the wire comprises a hypereutectoid pearlitic steel wire.

11. Wire according to claim 10, wherein the carbon content is between 0.85% and 0.95%.

12. Wire according to claim 10, wherein the diameter of the straight wire before the application of the deformations is between 130 µm and 175 µm and the tensile strength is between 3200 and 4500 MPa.

13. Method for producing a wire according to any one of claims 1 to 12.wherein a straight wire with a circular cross-section and a wire longitudinal axis is provided with a plurality of uniform pairs of plastic deformations in the form of displacements of the wire cross-section by passing it between the first teeth of a first pair of coplanar and meshing spur gears with parallel axes and subsequently passing it between the second teeth of a second pair of coplanar and meshing spur gears with parallel axes, wherein the axes of the first and second gear pairs are arranged perpendicular to each other, characterized in that the first pairs of displacements have a first wavelength ^1 and a first amplitude A1 and lie in a first plane containing the wire axis perpendicular to the wire axis,The second pairs of displacements have a second wavelength ^2 and a second amplitude A2 and lie in a second plane containing the wire axis perpendicular to the wire axis, the first and second planes are perpendicular to each other and each pair comprises exactly one protrusion and one indentation, and the first and second wavelengths are different from each other, ^1 ^ ^2, and the first and second amplitudes are exactly the same at the selected tensile stress of the wire sections in the wire creel, A1 = A2.

14. Method according to claim 13, wherein, prior to the application of the deformations, a multi-stage cold working is carried out by means of cold / wet drawing with a degree of deformation ^, Tbetween 3.40 and 3.

75.

15. Wire saw comprising two or more cylindrical wire guide rollers rotatably mounted on axes, the outer surfaces of which are provided with a plurality of uniformly spaced, annularly closed grooves in planes perpendicular to the respective axis, around which wire according to one of claims 1 to 12 is guided spirally in the grooves such that a flat gate facing the workpiece is stretched between two of the wire guide rollers, consisting of parallel sections of wire, wherein the wire sections are held under a selected tensile tension, and a feed device to which the workpiece is attached and which can be fed perpendicularly onto the wire gate, wherein the axes of the wire guide rollers and the workpiece are aligned parallel to each other.

16. Wire saw according to claim 15, wherein a tensile tension of the wire sections in the wire gate is between 60% and 70% of the tensile strength of the wire.

17. Method for simultaneously cutting a workpiece into a plurality of uniform slices using a wire saw according to claim 15 or claim 16, by feeding the workpiece towards the wire gate, rotating the wire guide rollers in the same direction so that the sections of wire in the wire gate perform a relative movement to the workpiece, and adding a slurry of silicon carbide (SiC) or diamond in a carrier fluid of glycol or oil, so that when the wire gate and workpiece are brought into contact, material is removed which, with further wire movement, feeding and supply of the slurry, cuts the workpiece into slices by means of a plurality of cutting edges.

Images