EP2396134A1

EP2396134A1 - Fixed abrasive sawing wire

Info

Publication number: EP2396134A1
Application number: EP10704355A
Authority: EP
Inventors: Glauber Campos; Xavier Amils; Davy Goossens
Original assignee: Bekaert NV SA
Current assignee: Bekaert NV SA
Priority date: 2009-02-13
Filing date: 2010-02-12
Publication date: 2011-12-21
Also published as: TW201043367A; WO2010092151A1; CN102292185A; SG173579A1; JP2012517906A; KR20110119705A

Abstract

A fixed abrasive sawing wire (40) is disclosed comprising a steel wire and abrasive particles (42) fixed thereon. The steel wire has a high carbon steel, high strength, pearlitic core (14) and a low carbon ferritic steel skin(12). The particles are indented in the skin (12) and further held in position by a binder layer(18). Good sawing results have been obtained which are conjectured to be due to the use of a low carbon (or even iron) skin(12). The fixed abrasive sawing wire (40) can conveniently be used on existing wire saws for cutting silicon wafers (solar and semiconductor) or other high cost materials, but without the need for an abrasive slurry.

Description

Fixed abrasive sawing wire

Description

Technical Field

[0001] The invention relates to a sawing wire, more specifically a sawing wire with fixed abrasive particles anchored in the outer skin of a carbon steel wire and fixed thereto with a binder layer. Specific about the wire is that a low carbon content steel is used as the skin on a high carbon content steel wire core. Such wires can be used for cutting hard and brittle materials like quartz (for e.g. quartz oscillators or mask blancs), silicon (for e.g. integrated circuit wafers or solar cells), gallium arsenide (for high frequency circuitry), silicon carbide or sapphire (e.g. for blue led substrates), rare earth magnetic alloys (e.g. for recording heads) or even natural or artificial stone.

Background Art

[0002] Wires for sawing large and small objects (called workpieces hereafter) made of hard and brittle materials are well known for centuries. In the field different names are used to denominate the member that actually contacts and saws the workpiece: the terms 'wire saw' (i.e. a saw that is a wire, although a saw apparatus using a wire could also be indicated by it), 'sawing wire' (a wire used for sawing) and a 'saw wire' (a wire used as a saw) all appear. In what follows, the term 'sawing wire' will be adhered to.

[0003] 'Sawing ropes' or 'sawing cables' are ropes comprising a rope made of several filaments on which abrasive beads are fixed which are sometimes also referred to as 'sawing wires' but this falls outside the scope of this application.

[0004] Basically the sawing wire serves as a carrier for an abrasive member that abrades materials from the object to be cut. These abrasive members: - can be separate from the carrier and injected by one or another means between the wire and the object to be sawn. The process is sometimes referred to as 'third body abrasion' (the third body being the abrasive member) or 'loose abrasive cutting'. A notorious example is the cutting of silicon ingots by means of a plain carbon steel wire that entrains slurry into the cut. The slurry contains fine abrasive particles that roll stick between the wire and the workpiece, crush the material locally and thereby further deepen the cut, or

- can be attached to the carrier wire in the form of protruding teeth made from the same material as the wire (as in a wood saw) or

- can be attached to the carrier wire in the form of abrasive particles of another material than the wire. In the latter case the particles must be hard and must be firmly attached to the carrier wire.

The interest of this application is in the last type of wire that will be called a

'fixed abrasive sawing wire'. What is expected from a fixed abrasive sawing wire is the following:

It must be supplied in sufficiently long lengths (at least kilometres) as short lengths necessitate a reciprocal motion. A reciprocal motion implies repetitive accelerations and decelerations hence loss of energy and time. The longer the wire, the less turnarounds are needed. In addition the thro and fro movement tends to wiggle the abrasive particles out of the wire leading to a premature wear of the wire due to loss of particles. So it is imperative that the abrasive particles are well fixed into or onto wire. With well fixed is not only meant that the abrasive particles must remain in place, but also that their elastic motion relative to the wire remains low.

- The thinner the wire the better. Thinner wires mean less kerf loss. The kerf loss is the amount of workpiece material that is abraded away and lost. Less kerf loss implies better use of material. For materials that are costly (such as silicon, gallium arsenide or rare earth magnet alloys) a small reduction in kerf loss results in a large financial gain. The benchmark is set in loose abrasive cutting where wires of a gauge 120 μm are customary and tests are underway with 80 μm wire. This results in a kerf loss of 130 to 140 μm and 90 to 100 μm as the abrasive particles in the slurry carrier also take some space between wire and workpiece. - As the wire must be tensioned during the sawing process in order to press the abrasive particles into the object, the wire must be able to sustain a certain tension. A higher tension that can be maintained results in more contact force on the abrasive particles and hence a higher sawing speed (although there is a limit to this). The tension is typically 20 N and higher.

- The desire for a low kerf loss i.e. a small width of the fixed abrasive sawing wire in combination with a high sawing speed requiring a high tension force, makes the use of high tensile strength wires indispensable. The tensile strength of the wire is defined as the breaking load - the force at which the wire breaks - divided by the cross-sectional area of the wire and is expressed in N/mm². In order to be on the safe side, the minimum breaking load of the wire should at least be about twice the tension force. In the case of e.g. a 200 μm fixed abrasive sawing wire this leads to a minimal tensile strength of 1300 N/mm² and in the case of a 140 μm to 2600 N/mm².

By preference the sawing wire should not pollute the workpiece with contaminants. One will for example try to eliminate copper in the sawing of silicon wafers not only because diffused copper in silicon is electronically active, but also because the waste treatment of copper containing effluent is more difficult.

[0006] Fixed abrasive sawing wires have been described in various patent applications and patents the most relevant for the purpose of this application being:

[0007] EP 0 243 825 describing a method to produce a fixed abrasive sawing wire starting from a steel wire rod and a tube surrounding the rod with a gap in between. The gap is filled with a mixture of metal powder and abrasive particles. The ends are sealed and the rod is heat treated and cold drawn in repeated steps to obtain a fixed abrasive sawing wire after the outer tube has been removed by etching it away. Drawbacks are that the method does not allow to produce fixed abrasive sawing wires of an appreciable length (above 100 meters), the tensile strength of the resulting wire is relatively low (say below 1800 N/mm²) and the resulting wires are too thick (1 mm).

[0008] EP 0 982 094 describes a fixed abrasive sawing wire with a stainless steel core, an intermediate layer for preventing hydrogen embrittlement of the core wire and a binding layer with diamond particles incorporated in them. The binding layer with the diamonds in it is deposited through electroplating or electroless deposition out of deposition bath comprising the diamonds. Embodiments given describe nickel as both the intermediate layer as well as the binding layer.

[0009] WO 99/46077 describes a fixed abrasive sawing wire comprising a metal wire, and superabrasive grains affixed to the wire through a brazed or soldered metal bond, wherein the grains are preferably disposed upon the surface with a preselected surface distribution. The application mentions that steel wires might lose strength due to the heat treatment needed for the brazing and soldering. This is not desirable to meet the tensile strength requirement.

[0010] EP 0 081 697 describes a method and an apparatus to incrust a wire with diamond particles. One departs from a wire that is coated with a copper or nickel layer prior to incrustation of diamond particles between hardened wheels that roll the wire around its axis through a repetitive axial movement of one or both of the wheels. Thereafter the diamonds are fixed in position by means of an electrolytically applied overcoat.

[0011] The abstract of JP 5016066 A2 describes the production of a sawing wire with a high carbon steel core and a low carbon steel skin through controlled decarburisation of a high carbon steel wire. However, the wire is intended for use with a loose abrasive slurry process. Hence the abrasive particles out of the loose abrasive slurry are not fixed in the carbon skin but get stuck and come loose again in a continuous way. Furthermore the decarburisation always results in a loss of carbon, hence a loss of strength of the wire. Disclosure of Invention

[0012] The main object of the invention is to provide a fixed abrasive sawing wire with improved properties. A further object of the invention is to provide a wire wherein the abrasive particles are not only well fixed but also show less elastic movement during sawing. Another object is to provide a fixed abrasive sawing wire that has a sufficiently high tensile strength to enable low kerf loss in combination with a high sawing speed. A method is described to make such wires in lengths longer than say one kilometer, which is a further object of the invention.

[0013] According to a first aspect of the invention a product in the form of a fixed abrasive sawing wire is disclosed. The sawing wire comprises a central wire made of steel. As steel is an alloy of iron and carbon and other elements it always comprises carbon in a certain amount. The outer periphery of the wire has a different composition than the inner core of the wire. In what follows this outer periphery will be called the skin. In the skin abrasive particles are fixed. A binder layer is applied on said skin to better hold the particles in the skin.

[0014] The cross section of the wire can have any suitable shape. The shape is dictated by the method of sawing. For example in multi-wire saws (wherein a single wire is repeatedly threaded in parallel over two or more guiding rolls, the name given to the machine is therefore somewhat misleading as only one wire is used) the cross section is preferably round. Indeed in such a multi wire saw the wire tends to rotate due to the many bendings over pulleys and guiding rolls, hence rotational symmetric wire i.e. round wire is most suited.

[0015] The overall diameter (i.e. including abrasive particles) of such a round fixed abrasive sawing wire can be from 80 micron up to 300 micron again determined by the machine it is used on. For example in sawing machines adapted to cut square blocks (12.5 x 12.5 cm²) out of one polysilicon block (of e.g. 1 by 1 meter) as described e.g. in patent CH 692489, a wire of around 250 micron would be more appropriate as a lot of force is needed to drag the wire through over the long length of 1 meter. In the case of lab sawing machines to cut small samples at low forces, an 80 micron wire might do the job. For slicing on multi-wire saws for the semiconductor industry, a wire of 100 to 200 micron seems most appropriate, where the user will of course favour the thinnest wire.

[0016] Alternatively the cross section can be of oval or even of rectangular shape. For example the tear drop shape as disclosed in US 5 438 973 is most preferred when using a frame saw (in a frame saw the individual wires are tensioned parallel to one another in a frame that is reciprocally moved over the workpiece). The tear drop shape allows to further reduce the kerf loss without giving in on strength. Also the high bending stiffness when bend in the plane of the longer side allows a higher sawing pressure in the cut.

[0017] Another alternative is to use flattened wires that are twisted around their own axis. Even star shaped cross sections could be useful in certain cases for example in hand operated saws.

[0018] The fixed abrasive steel wire is different from the prior art that the core of the steel wire has a pearlitic metallographic structure while the skin has a ferritic metallographic structure. The determination of the metallographic structure of a steel is a standardised technique: the wire is embedded into an epoxy block which is cut through perpendicular to the axis of the wire and subsequently polished. The shiny surface of the cross section is then etched in a nital solution which is a mixture of about 3% by volume nitric acid (HNO3) and alcohol, for example ethanol (C2H5OH). Due to the etching the grain structure of the steel becomes visible under a metallographic microscope at about 100 to 500 magnification.

[0019] Whenever in what follows a composition of a material in terms of its constituents in fractions of its total weight is given, it will be indicated as 'wt%' and should be read as 'percent by weight' unless it is made clear otherwise.

[0020] The pearlitic structure (or 'pearlite' for short) shows a brownish-grey pearly aspect (from which the name is derived) under the microscope. The pure pearlitic structure is formed after proper heat treatment of the steel (austinisation at temperatures above 723°C followed by slow cooling). Pearlite is a mixture of 88 wt% of ferrite (iron containing almost no carbon) and 12 wt% of cementite (Fβ3C) resulting in a eutectic concentration of 0.80 wt% carbon. When the carbon content of the steel is below 0.80 wt%, e.g. 0.40 wt%, steels are called hypo-eutectoid and formed pearlite is visible in regions that are surrounded by ferrite. When the carbon content is higher than 0.80 wt%, say 1.2 wt%, steels are called hyper-eutectoid and have a microstructure that comprises pearlite and cementite ('grain boundary cementite'). An experienced analyst can estimate the carbon content by weight through metallographic pictures in steps of about 0.2 wt% carbon.

[0021] As a pure pearlitic structure with exactly 0.80 wt% of carbon is a fiction, the main claim recites 'a substantially pearlitic metallographic structure'. For the purposes of this application a steel correctly treated to obtain such structure with a carbon content above 0.40 wt% is considered to show a 'substantially pearlitic metallographic structure'.

[0022] The skin of the fixed abrasive sawing wire shows a substantially ferritic metallographic structure or 'ferrite' for short. The ferrite is clearly discernable in a metallographic picture because it shows much lighter and is not coloured. For the purpose of this application ferrite is formed in steel with a carbon content of between 0.04 wt% and 0.20 wt%.

[0023] Practical steel compositions do not only comprise iron and carbon but a lot of other alloy and trace elements, some of which have a profound influence on the properties of the steel in terms of strength, ductility, formability, corrosion resistance and so on. As for this application strength is of the essence, the following elemental composition is preferred for the core of the steel wire:

- At least 0.70 wt% of carbon, the upper limit being dependent on the other alloying elements forming the wire (see below)

- A manganese content between 0.30 to 0.70 wt%. Manganese adds - like carbon - to the strain hardening of the wire and also acts as a deoxidiser in the manufacturing of the steel.

- A silicon content between 0.15 to 0.30 wt%. Silicon is used to deoxidise the steel during manufacturing. Like carbon it helps to increase the strain hardening of the steel. Presence of elements like aluminium, sulphur ( below 0.03%), phosphorous (below 0.30%) should be kept to a minimum. - The remainder of the steel is iron and other elements

[0024] The presence of chromium (0.005 to 0.30%wt), vanadium (0.005 to

0.30%wt), nickel (0.05-0.30%wt), molybdenum (0.05-0.25%wt) and boron traces may reduce the formation of grain boundary cementite for carbon contents above the eutectoid composition (0.80%wt C) and thereby improve the formability of the wire. Such alloying enables carbon contents of 0.90 to 1.20%wt, resulting in tensile strengths that can be higher as 4000 MPa in drawn wires. Such steels are more preferred and are presented in US 2005/0087270. If in the following reference is made to 'high carbon content' or 'high carbon steel', it is to be understood as the carbon content of the core of the steel wire.

[0025] The steel composition of the skin of the wire is less critical as it is predominantly iron with some carbon (between 0.04 wt% and 0.20 wt%) and other trace elements in it. In what follows when reference is made to 'low carbon content' or 'low carbon steel', it is to be understood as the carbon content or the steel of the skin of the steel wire.

[0026] All the above steel compositions are characteristic of a 'plain carbon steel' composition as the main alloying constituent is carbon. Steels that enable high strength are thus most preferred as the core of the wire must carry all force, the skin being of low strength, low carbon steel even further reduced in strength by the presence of the abrasive particles. Moreover - as in a circular cross section most of the area is at the periphery of the circle - a lot of area is low carbon steel hence does not contribute to the overall breaking load of the fixed abrasive sawing wire. This makes a fixed abrasive sawing wire of fine diameter with sufficient strength a non- obvious challenge.

[0027] Fixed abrasive sawing wires according to the invention typically have a tensile strength of above 2000 N/mm² for diameters smaller than 250 μm, above 2250 N/mm² for diameters smaller than 150 μm, and above 2500 N/mm² for diameter smaller than 120 μm. The tensile strength is defined as the breaking load of the fixed abrasive sawing wire divided by its metallic surface (excluding the area taken up by the abrasive particles). The metallic surface is determined on a cross section of the wire as used for the metallographic structure determination. Any metallic layer is taken into account for the surface. [0028] The local carbon content as radially measured from the core to the skin will show a decreasing function. This is schematically illustrated in Figures 2a or 2b where the radial local carbon distribution in percent by weight T(r)' is shown as a function of the distance from the centre 'r\ The steel wire has a radius 'R' and hence a diameter 2R. The average ' C ' carbon content can be found by integrating l^~(r) from '0' to 'R':

_ R /

C = JΪ(r) 2ττrdr /πR²

0 /

Experimentally, the easiest to determine is of course the average carbon content ' C ' e.g. by means of a LECO CS230 carbon and sulphur tester. The carbon content should be determined after removal of the particles and the fixing layer in order not to have interference of these. The average carbon content should at least be 0.40 percent by weight. More preferred is if it is above 0.55 wt% carbon or even above and 0.60 wt% carbon. [0029] The measurement of l^~(r) is difficult but can be done in a number of ways:

- There is the method of metallographic estimation. As mentioned before an experienced analyst can estimate the carbon content in classes about 0.2wt% carbon apart. In any case the presence of ferrite and pearlite is a standard procedure to determine.

- The most precise method is of course to locally obtain a microanalysis of the carbon content along the radius of the wire. Such analysis is currently possible in a scanning electron microscope equipped with a wavelength dispersive spectrometer (SEM-WDS). It is the ultimate method of reference.

[0030] An indirect measurement of the relative carbon distribution can be obtained through Vickers micro-hardness measurements. In this well known method (see ISO 6507-3 'Metallic Hardness Test: Vickers Test less than HV 0.2') a microindentor with a square-based diamond pyramid with a face angle of 136° is pushed with a specified force (0.9807 N, Hardness symbol HV 0.1) for a specified time (10 seconds) into the material. Thereafter, the geometry of the indentations is measured, out of which a local micro-hardness (expressed in N/mm²) can be calculated. In order to increase the spatial resolution of the method, a wire is encased in epoxy resin, cut under an angle to the axis of the wire and polished. The hardness at regular spots along the major axis of the ellipse that forms, is measured and the correct radial position with respect to the axis of the wire is calculated. The hardness measured is a function of the steel metallographic structure, the amount of strain hardening given to the wire (which is equal over the cross section of the wire) and the carbon content. The measurement of the Vickers micro-hardness is particularly important because it gives a measure how easy abrasive particles can be indented in the skin. [0031] In much the same way as with the carbon content, a weighed average

Vickers micro-hardness ΗV_aVg' can be calculated by replacing T(r)' by the local Vickers micro-hardness as a function of radial distance 'μ(r)':

HV_avg = R²

On experimental results, the integral can conveniently be approximated by taking the average of the under and upper sum of the discrete measuring points weighed with the annular surface between the points. [0032] By preference this average hardness is higher than 500, or more preferred between 550 and 650 N/mm². Too low an average hardness will not allow enough strength, too high an average hardness will not allow proper indentation of the particles. See Figures 2a and 2b for a schematic drawing. The skin can be defined as that part of the wire which has a below average hardness and the core as that part of the wire that has an above average hardness. The skin and the core meet at a border. At the border, the local Vickers micro-hardness crosses the weighed average micro-hardness. This border lies at an approximate radius 'b'. The thickness 'Δ' of the skin can then conveniently be defined as the radial distance between border and the outer perimeter of the wire or 'Δ = R-b'. [0033] The skin must prevent the core of the steel wire of micro-crack damage by indentation of the abrasive particles. Indeed, steel wires become more vulnerable to surface damage with increasing tensile strength. This is expressed in a loss of fatigue strength (as the damage is the start of a crack) and/or loss of strength. The skin must also hold the particles in position. Hence the indentation dept of the particles should never be larger than the skin thickness 'Δ\

[0034] According to a first preferred embodiment the transition from high carbon core to low carbon skin can be abrupt as shown in Figure 2a. Although there will be a carbon exchange between core and skin at nanoscopic level, no metallographic mixed phase is discernable on a microscopic level.

[0035] According to a second preferred embodiment the transition from high carbon core to low carbon skin is smooth and comprises a mixed metallographic phase showing increased ferrite presence and a decreasing pearlite presence when observing from core to skin. The transition becomes smooth due to the processing which will be explained later on. The carbon content distribution is then like the one depicted in Figure 2b. The width of the transition 'δ' can be defined as the distance between which μ(r) varies from 350 to 650 N/mm². The values correspond with what one can expect from a hard drawn, low carbon steel with less than 0.2 wt%C and a hard drawn, high carbon steel with more than 0.40 wt%C. The transition has as an advantage that the abrasive particle meets a constantly increasing indentation force, rather than an abrupt change when reaching the core of the wire when being pushed into the skin. The transition has also as an advantage that the skin is diffusion bonded to the core and loss of adhesion between skin and core is virtually impossible. The transition layer width 'δ' should therefore at least be wider than 5 micron, preferably wider than 10 micron in order to have an excellent bond.

[0036] At the other extreme, a too thick transition region will lead to an unacceptable loss of surface area for the high strength core and hence the wire will loose too much overall strength. Mutatis mutandis it will lead to a too hard outer skin. The width of the transition should therefore be kept below 40 micron, more preferably below 30 micron even more preferred below 20 micron.

[0037] Compared to conventional copper clad type of fixed abrasive sawing wires - as disclosed in EP 0 081 697 - the use of a low carbon cladding has the additional advantage that the low carbon cladding still adds to the overall strength of the wire. This is because low carbon has a higher cold deformation hardenability than copper. In this way finer fixed abrasive sawing wire can be made having the same strength as conventional sawing wires. This can reduce the kerf loss during sawing.

[0038] The abrasive particles can be superabrasive particles such as diamond (natural or artificial, the latter being somewhat more preferred because of their lower cost and their grain friability), cubic boron nitride or mixtures thereof. For less demanding applications particles such as tungsten carbide (WC), silicon carbide (SiC), aluminium oxide (AI2O3) or silicon nitride (Si3N₄) can be used: although they are softer, they are considerably cheaper than diamond. Most preferred is diamond.

[0039] The size of the abrasive particles somewhat scales with the diameter of the wire. Determining the size and shape of the particles themselves is a technical field in its own right. As the particles have not - and should not have - a spherical shape, for the purpose of this application reference will be made to the 'size' of the particles rather than their 'diameter' (as a diameter implies a spherical shape). The size of a particle is a linear measure (expressed in micrometer) determined by any measuring method known in the field and is always somewhere in between the length of the line connecting the two points on the particle surface farthest away and the length of the line connecting the two points on the particle surface closest to one another.

[0040] The size of particles envisaged for the fixed abrasive sawing wire fall into the category of 'microgrits'. The size of microgrits can not longer be determined by standard sieving techniques which are customary for macrogrits. In stead they must be determined by other techniques such as laser diffraction, direct microscopy, electrical resistance or photosedimentation. The standard ANSI B74.20-2004 goes into more detail on these methods. For the purpose of this application when reference is made to a particle size, the particle size as determined by the laser diffraction method (or 'Low Angle Laser Light Scattering' as it is also called) is meant. The output of such a procedure is a cumulative or differential particle size distribution with a median dso size (i.e. half of the particles are smaller than this size and half of the particles are larger than this size) or in general 'dp' wherein 'P' percent of the particles is smaller than this 'dp' the remaining part (100-P) percent being larger sized than

[0041] Superabrasives are normally identified in size ranges by this standard rather than by sieve number. E.g. particle distributions in the 20-30 micron class have 90% of the particles between 20 micrometer (i.e. 'ds') and 30 micrometer (i.e. 'dθs') and less than in 1 in 1000 over 40 microns while the median size dso must be between 25.0 +/ 2.5 micron.

[0042] As a rule of thumb, the median size (i.e. that size of particles where half of the diameters have a smaller size and the other half a larger size), should be smaller than 1/12^th of the circumference of the steel wire, more preferably should be smaller than 1/18^th the circumference of the steel wire in order to accommodate the particles well in the skin. At the other extreme the particles can not be too small as then the material removal rate (i.e. the amount of material abraded away per time unit) becomes too low.

[0043] The abrasive particles will be best held when they are indented into the skin over more than half their median size. So the indentation depth should at least be larger than half the median size of the particles. As the skin thickness should be thicker than the largest indentation depth, it follows that the skin thickness must at least be larger than half the median particle size in order to hold the particles well.

[0044] Even more preferred is if the skin thickness is thicker than 'dgo' (90% of the particles have a size that is smaller than dθo). Hence the chances for micro-crack damages to the core become very small thereby avoiding breaks in service. In practise the skin thickness will have to be about 10% of the diameter of the steel wire and should be at least 5% or at least be 7% of the diameter. So 8 μm to 15 μm for very thin wires of 80 to 150 μm and about 20 μm for a 210 μm wire. Note that with a 10% of the diameter skin thickness, already 36% of the cross sectional area of the wire is occupied by low tensile skin material.

[0045] Whether or not the particles are properly indented can be readily assessed on a cross section of wire wherein the abrasive particles are removed and hence crevices remain in the skin. At an indentation such a cross section will reveal two extreme points at the outer circumference of the wire that can be connected by a line. The distance from point to point will be called the indentation width 'W_1n'. The largest distance taken perpendicular to the line reaching the steel wire will be called the indentation depth 'D_1n'. In order to be well fixed, twice the indentation depth must be larger than the indentation width. When a lot of crevices are measured, good indentation is characterised by an average of the ratios (2xDι_n/Wι_n) that is larger than one. At least 20 crevices should be measured.

[0046] As to how many particles must be present at the surface of the sawing wire, much depends on the type of material to be cut. A too high density will induce too low forces on the particles which will polish the particles resulting in a decrease of their cutting ability. On the other hand a too low density may lead to particles being torn out of the skin as the forces become too large or to too low cutting rate as not enough particles pass the material per unit time. The presence of particles can be quantified by the ratio of the area occupied by the particles to the total circumferential area of the wire: the 'coverage ratio'. This can be done by SEM by selecting the particles with a typical composition out of the general picture and calculating the occupied area by the particles relative to the total area. Only the centre part of the wire picture should be used as the sides tend to overestimate the particle surface due to the turning away of the wire surface. An example is given in Figure 4.

[0047] The target coverage ratio for the particles is depending on the material one intends to cut, the cutting speed one wants to reach or the surface finish one wants to obtain. The inventors have found that in order to have the best sawing performance for the materials envisaged the ratio of particle area over total area should be between 1 and 50%, or between 2 to 20% or even between 2 and 10%.

[0048] The binder layer that is applied onto the outer surface of the wire and helps to keep the particles fixed in the skin or in other words to bind the particles in the skin. By preference the binding layer is a metallic layer. Particularly favoured metals are nickel and iron. Alternative but still preferred metals are chromium, cobalt, molybdenum, tungstenand zinc and alloys thereof. The thickness of this layer is preferably between 1 to 5 μm.

[0049] From the above it is clear that the inventive fixed abrasive sawing wire is substantially free of copper. No intentionally added copper is present in the wire or in the coating. Hence, contamination with copper of silicon workpieces is avoided during drawing. Copper atoms that diffuse into silicon form electronically active defects in the energy gap of silicon. Also the elimination of copper out of effluent streams (e.g. resulting from coolant, or rinsing of the finished wafers) can be avoided in this way.

[0050] According to a second aspect of the invention, a method to produce the wire is disclosed. In general such method comprises three main steps:

> First, one has to provide a steel wire that has a high carbon core and a lower carbon skin of sufficiently fine diameter.

> Second, one indents the skin with the abrasive particles of the preferred size and type.

> In a third step the abrasive particles are fixed with a binding layer.

By preference the second and third step are implemented in a line concept where the wire is continuously fed from one process step to the following process step. However, separation of these steps is not excluded by that: e.g. a batch process as described in EP 1375043 for the third step is possible. There are a number of ways in which the starting steel wire of the first step can be produced:

[0051] In a first embodiment of the first method step, a high carbon steel wire is coated with pure iron from an electrolytic bath (see for example US 5014760). Some alternative approaches become possible. [0052] A first alternative is that the final diameter steel wire is coated with iron. The transition between the iron skin layer and the high carbon core is abrupt and no mixed phase between core and skin will form. The embodiment as described in Figure 2a is then obtained. The advantage of this method is that relatively little iron must be laid down on the wire to reach a reasonable layer thickness (thickness more than 7% of steel wire diameter).

[0053] A second alternative is that, the steel wire can be coated with iron at a suitable intermediate diameter prior to further wet wire drawing. With an intermediate diameter is meant a diameter between the wire rod diameter and the final diameter of the wire (an intermediate diameter will typically lay between 2.70 and 0.90 mm). The heat generated during drawing will result in the formation of a minor transition region of about 5 micron or more due to the diffusion of the carbon into the iron.

[0054] In a third alternative the wire can be coated with iron at an intermediate diameter level and - possibly repeatedly - patented and drawn. In this case the transition region is higher due to the single thermal treatment of the skin which brings more diffusion of the carbon into the iron. The transition region is then between 5 and 30 micron.

[0055] According to a second preferred embodiment of the first method step, the high carbon steel core is wrapped with a low carbon steel strip or iron foil that is closed by welding and forms the skin. Again - like in the first preferred embodiment - alternatives are possible:

[0056] In a first alternative an intermediate steel wire of diameter 2.40 to 0.90 mm is wrapped with an iron foil or low carbon strip prior to further wet wire. Again, the heat generated during drawing will result in the formation of a minor transition region of about 5 micron due to the diffusion of the carbon into the iron.

[0057] In a second alternative the high carbon steel core is wrap-coated with a low carbon or iron strip or foil at an intermediate diameter level and subsequently - possibly repeatedly - patented and drawn to final diameter. Again the transition region is somewhat broader due to the single or possibly two or three thermal treatment(s) of the wire inducing more diffusion of the carbon into the skin. The transition region is then between 5 and 30 micron. The transition region increases with the number of patenting steps.

[0058] The first and second preferred embodiments of the first method step result in a hardness profile that tails up at the surface of the skin rather than showing a continuous decrease.

[0059] According to a third preferred embodiment of the first method step, the skin is formed by decarburisation of a high carbon steel wire. Practical examples of decarburization are given in US 5014760. The outer layer of the steel wire then loses a substantial part of its carbon and forms a low carbon skin while the core retains most of the carbon. As decarburisation requires passing the wire at elevated temperatures of 700°C to 1000°C in an oxidizing atmosphere furnace, it is not possible to decarburize the final diameter wire as this would lead to unacceptable strength loss.

[0060] Hence decarburization is by preference performed on intermediate wire diameters of higher than about 0.90 mm. The decarburization step can be performed on rod diameter level and followed by one or two regular (i.e. under reducing atmosphere) patenting steps with wire drawing operations in between and after. Alternatively, the decarburization step can be the last thermal treatment prior to final wire drawing. The latter is somewhat more preferred, as a subsequent regular patenting (under reducing atmosphere) results in a redistribution of carbon in the wire. Such redistribution results in a broadening of the transition region.

[0061] In general the decarburization of a high carbon steel wire inevitably leads to loss of carbon i.e. loss of overall strength which works against the requirement of the need for strength. In order to reach the needed overall strength on the final product, the carbon content of the high carbon starting wire may have to be prohibitively high. Over and above the out diffusion of carbon leads to a carbon profile that my result in a too soft outer layer. Hence the hardness profile will steadily diminish and not show a tailing up at the skin. The carbon profile is also difficult to control as the skin thickness is dictated by the diffusion law. [0062] During the second method step the skin of the wire is indented with abrasive particles. This can conveniently be done by temporarily fixing the abrasive particles to the wire prior to rolling them into to the skin by means of rolls. An example how this can be done is disclosed in EP 008169. Improvements to that art are e.g. to temporarily fix the particles by applying a viscous substance in which the particles stick that later on can be washed away (preferably in water). A further improvement is that the rolling is done between hardened rolls with matching semicircular grooves through which the wire is led. Another improvement is that different pairs of rolls under different angles can follow one after the other.

[0063] In the third process step, the particles are fixed by means of fixing layer that is by preference metallic in nature. Application of the fixing layer should be done under low temperature conditions (below about 200°C) in order to avoid tensile strength degradation of the wire. The most preferred method is therefore to use an electrolytic deposition technique to deposit metal ions out of a metal salt electrolyte onto the wire that is held at a negative potential relative to the electrolyte. Even then care has to be taken not to have excessive resistive heating of the steel wire as steel is a less good electrical conductor and the wire is fine. Also the presence of the particles makes making the electrical contact to the wire difficult as the particles are insulators by nature and a simple rolling contact will result in sparking. Hence a non-contact method as e.g. described in WO 2007/147818 is preferred wherein contact with the wire is made through a second electrolyte in a bath separated from the metal deposition electrolyte bath.

Brief Description of Figures in the Drawings

[0064] Fig. 1 'a' and 'b' are different metallographic cross sections of the same wire where the indented particles have been removed out of their indentation. [0065] Fig. 2 'a' schematically depicts the radial concentration by weight of the carbon content l^~(r) or the local microscopic Vickers hardness μ(r) in the case of steel wire with an abrupt transition from high carbon core to low carbon skin. [0066] Fig. 2 'b' schematically depicts the radial concentration by weight of the carbon content l^~(r) or the local microscopic Vickers hardness μ(r) in the case of steel wire in the case of steel wire with a smooth transition from high carbon core to low carbon skin. [0067] Fig. 3 'a' shows an actual measurement of the local microscopic Vickers hardness of a first example. [0068] Fig. 3 'b' shows an actual measurement of the local microscopic Vickers hardness of a second example. [0069] Fig. 4 shows how the coverage percentage of the abrasive particles can be determined. [0070] Fig. 5 shows how the indentation width and depth of a particle can be measured.

Mode(s) for Carrying Out the Invention

[0071] According to an example of the invention, a high carbon wire rod (nominal diameter 5.5 mm) with a carbon content of 0.8247 wt%, a manganese content of 0.53 wt%, a silicon content of 0.20 wt% and with Al, P and S contents below 0.01 wt% was chemically descaled to the methods known in the art.

[0072] The wire was subsequently wrapped with a low carbon strip with 0.03 wt% carbon and a thickness of 0.60 mm. The seam was welded. The total diameter of the wrapped wire was thus about 6.7 mm. The strip thickness is 8.96% of the total wire thickness. [0073] This composite wire was dry drawn in the manner known in the art to a total diameter (i.e. core wire plus strip wrap) of 2.40 mm. The material was split in two separate batches.

[0074] A first batch of material (referred to as example 1) was further dry drawn to a total diameter of 1.20 mm. The thickness of the low carbon strip was thereby reduced to 105 μm (i.e. 8.75 % of the total wire thickness). This material was then patented in the usual manner (lead patenting). After patenting there is already clear indication of carburisation of the low carbon strip and the strip is fully fused to the core. Thereafter, another dry drawing step to 0.90 mm total diameter is performed. This wire was subsequently wet wire drawn to a total diameter of 210 μm. Due to the patenting the low carbon strip was carburised and the transition from core to skin was not longer clearly discernable. The sample has undergone only one patenting operation.

[0075] A second batch of material (referred to as example 2) was first patented in lead and subsequently dry drawn to 0.90 mm diameter and again patented in lead. Thereafter it was wet wire drawn to again 210 μm. This sample has undergone two patenting operations.

[0076] A comparison between both samples is given in Table I:

Table I

'Skin μHV refers to the measured Vickers micro-hardness as measured on the drawn wire. [0077] The initial Vickers micro-hardness of the low carbon steel strip was 143 N/mm². This appears to have considerably increased due to:

> The high degree of conforming during drawing which is known to result in a harder material.

> The carburization of the skin material: a higher carbon content is known to result in a harder material.

It appears that the carburisation plays the more dominant role for the hardening of the skin as example 1 has undergone a larger reduction but only one patenting step while the example 2 has obtained a lower final reduction while being patented twice.

[0078] The hardness profile of the wire of example 1 was measured and is represented in Figure 3a. The hardness was measured on an elliptical cross section such that the respective indentations were sufficiently far apart (indicated with '♦'). The outer point ('■') is the point measured on the outer skin (cfr. table I). The dash-dot line marked with ΗV_aVg' indicates the average micro-hardness weighed with the surface area and in this case was equal to 597 N/mm² (lower sum 586, upper sum 606). The thickness TY of the skin is that distance from the outer circumference to where the hardness is equal to the average Vickers micro-hardness. In this case the border between skin and core is between 80 and 84 μm radius, hence the skin thickness TY is about 21 to 25 μm. The skin is thus about 8.5 to 12% of the steel wire diameter. The transition region 'δ' is about 17 μm thick.

[0079] The hardness profile of the example 2 wire is represented in Figure 2b.

The different symbols ('♦' and '*') represent a repeated measurement. The average weighed micro-hardness was respectively 577 N/mm² ('*', indicated with the dash-dot line; lower sum 559, upper sum 595) and 589 N/mm² ('♦', indicated with the dash-dot-dot line, lower sum 571 , upper sum 607). The skin thickness 'Δ' is about 22 μm while the transition region 'δ' is broader namely 23 μm.

[0080] Comparing the hardness curves of example 1 and 2 clearly demonstrates that the double patenting results in a harder overall skin hardness although the hardness at the surface does not differ that much. Clearly, carbon has evolved from the high carbon core to the low carbon skin. [0081] Example 2 was selected for further processing. The steel wire of example 1 was indented with diamond particles of median size 'dso' 25.3 μm (dio=15.1 μm, d90=40.6 μm). This was achieved by dip-coating the wire with a water soluble adhesive (available from Aquabond™). Immediately thereafter the wire with particles is led in between two pairs of rolls with a matching semi circular groove of radius 109 μm. The two pairs had their axes perpendicular to one another.

[0082] In a subsequent deposition, the wire was coated with a nickel binding layer after washing the adhesive away in hot water. This was done in an installation as described in WO 2007/147818. The thickness of the layer was about 3 micron.

[0083] The coverage degree of the wire was about 5 to 8 % and was determined in a SEM in backscattered electron detection mode. Figure 4 shows the resulting picture with diamond particles on the surface of the fixed abrasive sawing wire 40 as black areas 42 on an otherwise grey background 44. By means of photoanalysis software the ratio black area over black and grey area or coverage degree can be readily assessed.

[0084] Metallographic cross sections are shown in Figure 1a and 1 b which are cross sections of the same wire 10, but on different places. When looking through binoculars, it is clear that the core 14 of the wire 10 shows a different structure than the skin 12. The core 14 shows a high carbon, drawn pearlitic metallographic structure while the skin 12 shows a substantially ferritic structure i.e. with a low carbon content. The originally circular cross section of the wire has been indented with particles which are subsequently drawn out during polishing of the sample. They leave an indentation 16. It is clear that the indentation occurred prior to the coating with the nickel binding layer 18 as no nickel layer is visible inside the crevice left by the diamond. The crevice is about 20 micron deep (measured from the outer nickel surface) and does not enter the core.

[0085] That the indentation of the particles did not adversely affect the strength of the sawing wire was ascertained in a breaking load test of the finished wire: no noticeable loss in breaking load was observed compared to the results obtained on the steel wire (table I). Hence, the indentation did not damage the high strength steel core.

[0086] The quality of the indentation can be estimated by comparing the width of the crevice to its depth. How this can be done is illustrated in Figure 5 wherein a cross section 50 as in Fig. Va' or 'b' is reproduced. When connecting the outer points 'A' and 'B' of the crevice 52, the width 'Wm' can be determined. Likewise the depth 'D_1n' is determined by measuring the maximum depth perpendicular to the line AB. The measure (2xDι_n/Wι_n) is independent of where exactly the cross section has been taken.

[0087] To this end the width and depth of 20 indentations was measured on cross-sections perpendicular to the wire axis like the ones depicted in Figure 1 (although longitudinal cross-sections are equally well suited). The minimum ratio (2xDι_n/Wι_n) was 0.45 and maximum ratio was 2.57, the average was 1.17 which is larger than 1 meaning that the indentation is sufficient.

[0088] The performance of the fixed abrasive sawing wire was confirmed on a Diamond Wire Technology CT800 reciprocal lab saw machine. Half of a single crystal silicon semi-square of 10 cm width and 5 cm height was cut several times by the inventive wire. The machine was operated in 'constant bow mode' set at 3°, the wire tension was kept constant at about 15 N, 30 m of wire was cycled (thro and fro) in 7 seconds giving an average speed of (2x30/7=) about 8.6 m/s. Water with an additive was used as a coolant.

[0089] In table Il a comparison is given of the sawing speed as disclosed in the prior-art (only silicon samples have been considered) with that obtained by the inventive sample: [0090]

Table Il [0091] In another series of experiments, the gain in strength by using a low carbon clad wire in stead of using another kind of cladding material such as copper was assessed. Starting from a 0.80% carbon steel two 0.30 mm wire samples were made: one with a low carbon cladding and one with a electrolytic copper cladding. The resulting wires are characterised by the data in Table III:

Table

The difference in core tensile strength and steel cross sectional area can account for 18 N (= 146-128 N) of the observed difference in breaking load of 43 N (188 - 145 N). The remaining difference of 25 N (=43-18 N) can only be attributed to the difference in material low carbon versus copper. Replacing copper with low carbon in a fixed abrasive sawing wire can lead to an increase of 17 % in breaking load, keeping all other things equal. It is thus possible to reduce further the diameter of the fixed abrasive wire - and thus kerf loss - while keeping the same breaking load.

[0092] The low carbon clad wire was further indented with diamond particles and these particles were fixed by means of a nickel layer. Two different degrees of coverage degree were made: one with about 0.60% coverage and one with about 2% coverage degree. The samples were tested on a piece of mono crystalline silicon according to the same protocol as described before (paragraph [0088]) but with a variation in tension. For the 2% coverage ratio, at 18 N tension, a sawing speed of 133 mm²/min was obtained which increased to 164 mnrVmin under a tension of 27 N. The 0.60% coverage ratio sample showed inferior cutting results.

[0093] Hence, much to the surprise of the inventors, the use of low carbon steel for a skin of a fixed abrasive sawing wire turned out to be a very good choice for holding the abrasive particles. While in the prior-art other metals are used on top of the steel core as skin (like copper and nickel) the use of low carbon steel as the skin material turned out to have many advantages which are hypothised to be related to the skin material:

- The modulus of elasticity of iron is 220 000 MPa compared to 124 000 MPa for copper and 196 000 MPa for nickel. Hence when the abrasive particle is wiggled thro and fro in the sawing process, the iron will give a stronger support to the particle than e.g. copper.

- The skin material adheres very well to the core material. When low carbon steel is put on high carbon steel the materials are compatible, hence adhere better to one another.

Moreover, when some diffusion of carbon can take place between the high carbon core and the skin, the skin and the core are as if welded to one another.

- When a wire is repeatedly mechanically loaded (by bending or tensioning) in a corrosive environment wherein hydrogen forms the wire will prematurely break which is known as hydrogen induced corrosion cracking. The phenomenon occurs because the hydrogen enters the steel structure and makes the steel brittle (hydrogen embrittlement). For example a fixed abrasive sawing wire which is constantly flooded with a coolant liquid is prone to such corrosion. A closed ferritic layer is known to reduce the hydrogen induced corrosion cracking of a wire as is explained in US 5 014 760.

- It is further conjectured that, in the case the particles are diamond particles, the carbon of the diamond will at least diffuse somehow into the iron, thereby hardening the indented low carbon steel in which the diamond is embedded. Furthermore such a diffusion may lead to a better adhesion of the diamonds in the skin. These hypothesis were formulated after the tests described and are an attempt to explain the remarkable results and should not be used to render the invention obvious as they are mere hypothesis formulated after the facts.

Claims

1. A sawing wire comprising a steel wire, said steel wire having a core and a skin, abrasive particles fixed in said skin and a binder layer on said skin for binding said particles in said skin characterised in that said skin has a substantially ferritic metallographic structure for enabling indentation of said particles while said core has a substantially pearlitic metallographic structure for providing strength to said sawing wire.

2. The sawing wire according to claim 1 wherein said steel wire has an weighed average Vickers micro-hardness, said skin having a local Vickers micro- hardness lower than said average, said core having a local Vickers micro- hardness higher than said average, said core and said skin meeting at a border.

3. The sawing wire according to claim 2 wherein said weighed average Vickers micro-hardness is at least 500 N/mm².

4. The sawing wire according to any one of claims 1 to 3 wherein the transition from said core to said skin is abrupt at said border and no mixed phase is metallographically discernable.

5. The sawing wire according to any one of claims 1 to 3 wherein the transition from said core to said skin at said border is smooth and comprises a mixed metallographic phase of ferrite and pearlite.

6. The sawing wire of claim 5 wherein the width of said transition is larger than 5 micron, said width being defined as the radial distance wherein the local Vickers micro-hardness varies from 400 to 650 N/mm².

7. The sawing wire according to any one of claims 1 to 6 wherein said steel wire has an average carbon content of at least 0.40 percent carbon by weight.

8. The sawing wire according to claim 7 wherein the overall tensile strength of the wire is larger than 2 000 N/mm².

9. The sawing wire according to any one of claims 2 to 8 wherein said skin has a skin thickness defined as the radial distance measured from the outer perimeter towards said border, said particles having an indentation depth relative to outer perimeter of said wire, said skin thickness being at least the largest indentation depth to avoid damage to said core.

10. The sawing wire according to claim 9 where said skin thickness is at least dθo for improved anchoring of said particles in said skin, wherein dθo is that size of particles for which 90 out of 100 particles have a size below dθo.

11. The sawing wire according to claim 10 wherein said skin thickness is at least half the median size of said particles.

12. The sawing wire according to any one of claims 9 to 11 wherein said steel wire has a diameter and said skin thickness is more than 7% of said diameter.

13. The sawing wire according to any one of claims 1 to 12 wherein the fraction of particle covered area divided by total wire area is between 1 and 50 percent.

14. The sawing wire according to any one of claims 1 to 13 wherein said binder layer is a metallic binder layer the metal being one out of the group comprising iron, nickel, chromium, cobalt, molybdenum, tungsten, cupper, zinc and alloys thereof.

15. The sawing wire according to any one of claims 1 to 14 wherein said abrasive particles are selected out of the group comprising diamond, cubic boron nitride, silicon carbide, aluminium oxide, silicon nitride, tungsten carbide or mixtures thereof.