DE102006062872B4 - Simultaneous double-side grinding providing method for semiconductor wafer, involves machining semiconductor wafers in material-removing fashion between rotating upper working disk and lower working disk - Google Patents

Abstract

The method involves machining semiconductor wafers (15) in material-removing fashion between two rotating upper and lower working disks (1, 4), where the working disks comprises working layers (11, 12) containing bonded abrasive. The temperature in the working gap (30) is kept constant during the machining, where each semiconductor wafer lies such that the wafer is freely moveable in a cutout of one of a set of carriers caused to rotate by a rolling apparatus. An independent claim is also included for a semiconductor wafer comprising an isotropic ground pattern.

Description

The subject of the present invention is a method for simultaneous double-sided grinding of a plurality of semiconductor wafers, each wafers being freely movable in a recess of one of a plurality of wafers rotated by a winder and thereby moved on a cycloidal trajectory, the wafers between two rotating working wafers eroding, wherein each working disk comprises a working layer containing bonded abrasive.

State of the art

For electronics, microelectronics, and microelectromechanics, the starting materials (substrates) required are semiconductor wafers with extreme requirements for global and local flatness, single-sided local flatness (nanotopology), roughness, and cleanliness. Semiconductor wafers are wafers of semiconductor materials, in particular compound semiconductors such as gallium arsenide and predominantly elemental semiconductors such as silicon and occasionally germanium. Optionally, layer structures are first produced on the semiconductor wafers before they are used for the production of components. Layer structures are z. A silicon-on-insulator (SOI) or strained silicon-germanium layer on a silicon wafer or combinations of both (strained silicon-on-insulator, sSOI) ,

• a) Herstellung eines einkristallinen Halbleiterstabs (Kristallzucht);
• b) Auftrennen des Stabs in einzelne Scheiben;
• c) mechanische Bearbeitung;
• d) chemische Bearbeitung;
• e) chemo-mechanische Bearbeitung;
• f) ggf. Herstellung von Schichtstrukturen.

According to the prior art, semiconductor wafers are produced in a multiplicity of successive process steps, which can generally be classified into the following groups:

• a) production of a monocrystalline semiconductor rod (crystal growth);
• b) separating the rod into individual slices;
• c) mechanical processing;
• d) chemical processing;
• e) chemo-mechanical processing;
• f) if necessary, production of layered structures.

The combination of the individual steps attributable to the groups and their order varies depending on the application. Furthermore, a variety of minor steps such as cleaning, sorting, measuring, packaging, etc. is used.

The mechanical processing is used to remove ripples that have arisen during the previous separation of the semiconductor rod, for example by thermal drift over the long separation time or dynamic self-sharpening and Abstumpfungsprozesse. Furthermore, the mechanical processing is used for removing the surface layer which has been damaged in crystalline form by the rough sawing process and for reducing the surface roughness. Above all, however, mechanical processing is used to globally planarize the semiconductor wafer. According to the state of the art, different techniques are used here, such as lapping (double-side lapping with free abrasive grain), single-side grinding with a single-side grinding wheel (SSG), or simultaneous double-side grinding between two cup grinding wheels. and backside at the same time ("double-disk grinding", DDG).

DE10344602A1 describes a method that combines the lapping-known kinematics and constraining-force-free guidance with the advantages of the bonded abrasive grain. In this case, the semiconductor wafers are usually moved with a plurality of carriers between an upper and a lower working disk. The two work disks are, for example, covered with a sanding cloth. The rotor discs, each having a plurality of recesses for receiving the semiconductor wafers are like a lapping machine via a sprocket with a rolling device, consisting of an inner and an outer drive ring, in engagement and through this in a rotating movement about its axis and order the axis of the drive rings offset, so that describe the semiconductor wafers relative to the also about its axis rotating working disks cycloid traction. DE19626396A1 describes a double side grinding process with the same kinematics. The workpieces are processed between two rotating grindstones with the addition of a grinding fluid.

Th. Ardelt, Reports from the Production Technical Center Berlin, Fraunhofer Institute for Production Systems and Design Technology, IPK Berlin, 2001, ISBN 3-8167-5609-3, describes the influence of the relative movement between workpieces and working wheels on the result of the surface grinding Planet kinematics and provides detailed information on the analysis of the planetary kinematics and the simulation of the grinding wheel load.

US2003 / 0054650A1 goes into detail about the kinematics of this process and states what needs to be met by the trajectories in order to achieve an optimal result in terms of flatness and freedom from scratches in a double-side polishing. These requirements apply accordingly US2003 / 0054650A1 also for double side grinding.

However, it has been found that the wafers processed by this method have a number of defects, so that the resulting wafers are unsuitable for particularly demanding applications: For example, it has been found that semiconductor wafers with unfavorable spherical thickness profile generally result in pronounced edge drop , The semiconductor wafers often also have irregular ripples in their thickness profile and a rough surface with great depth of damage. By depth of damage is meant the depth, calculated from the surface of the semiconductor wafer, up to which the crystal lattice is damaged by the processing, ie. H. was disturbed.

Rough semiconductor wafers with great depth of damage require a complex post-processing, through which the advantages of in DE10344602A1 disclosed method be nullified. Due to the usual chemical and chemo-mechanical follow-up processing, spherical semiconductor wafers can hardly be converted into the desired plane-parallel target shape, or only with great effort. The remaining crowning and the remaining edge drop leads to incorrect exposures during the photolithographic component structuring and thus to the failure of the components. Such semiconductor wafers are therefore unsuitable for demanding applications.

task

The object of the present invention is therefore to provide semiconductor wafers which, due to their geometry, are also suitable for the production of electronic components with very narrow linewidths ("design rules").

Furthermore, the task was to avoid the formation of a Randabfalls in the production of semiconductor wafers.

It was also an object to avoid other geometric errors such as a thickness maximum in the center of the semiconductor wafer associated with a steadily decreasing thickness towards the edge of the disk or a local minimum thickness in the center of the semiconductor wafer.

solution

The object is achieved by a first method for simultaneous double-sided grinding of a plurality of semiconductor wafers (which, however, is not the subject of the invention), each semiconductor wafer being freely movable in a recess of one of a plurality of carrier discs set in rotation by means of a winder, thereby moving on a cycloidal trajectory is, wherein the semiconductor wafers are machined material between two rotating working discs, each working disc comprises a working layer containing bonded abrasive, characterized in that during the processing, the temperature prevailing in the working gap is kept constant.

The object is likewise achieved by a second method for simultaneous double-sided grinding of a plurality of semiconductor wafers (which is likewise not the subject of the invention), wherein each semiconductor wafer is freely movable in a recess of one of a plurality of winder wheels rotated by means of a winder and thereby a cycloidal trajectory is moved, wherein the semiconductor wafers are machined material between two rotating working discs, each working disc comprises a working layer containing bonded abrasive, characterized in that per unit time the amount of the number of revolutions of the carriers around the center of the rolling device and relative to each of the two working wheels is greater than the amount of the number of revolutions of the individual carriers around their respective centers.

der beiden Arbeitsschichten zum Mittelwert der Beträge der Abnutzung beider Arbeitsschichten für jede radiale Position r weniger als 1/1000 beträgt, wobei der Betrag der theoretischen Abnutzung einer jeden Arbeitsschicht gegeben ist durch

wobei a den Teilkreisradius der Umlaufbewegung der Läuferscheiben auf den Arbeitsscheiben um den Mittelpunkt der Abwälzvorrichtung; e den Abstand des momentan betrachteten Aufpunkts vom Mittelpunkt der entsprechenden Läuferscheibe; l(e) die innerhalb der Fläche der Halbleiterscheibe verlaufende Bogenlänge des Kreises mit Radius e um den Mittelpunkt der entsprechenden Läuferscheibe; r die radiale Position bezüglich des Mittelpunkts der Arbeitsscheiben; σ_i die Winkelgeschwindigkeit des Umlaufs der Läuferscheiben um den Mittelpunkt der Arbeitsscheiben; ω_i die Winkelgeschwindigkeit der Eigendrehung der Läuferscheiben um ihre jeweiligen Mittelpunkte angibt, e_min = max{0; e_exz – R} und e_max = e_exz + R mit R = Radius der Halbleiterscheibe die untere und obere Grenze der Integration über e bezeichnen; e_exz die Exzentrizität der Halbleiterscheibe in der Läuferscheibe und der Index i = o für die obere Arbeitsscheibe oder i = u für die untere Arbeitsscheibe angibt, ob die Winkelgeschwindigkeiten σ_i und ω_i sich auf die obere oder die untere Arbeitsscheibe beziehen.The object is achieved by a third method for simultaneous two-sided grinding a plurality of semiconductor wafers, each wafer is freely movable in a recess of a plurality of offset by means of a rolling device rotors and is thereby moved on a cycloidal trajectory, wherein the semiconductor wafers between two rotating Working discs are machined material, each working disc comprises a working layer containing bonded abrasive, characterized in that the angular velocities σ _i and ω _i are chosen so that the amount of the ratio of the difference of the amounts of theoretical wear

of the two working layers to the mean of the amounts of wear of both working layers for each radial position r is less than 1/1000, wherein the amount of theoretical wear of each working layer is given by

where a is the pitch radius of the orbital movement of the carriers on the working wheels around the center of the rolling device; e is the distance of the currently considered point from the center of the corresponding rotor; l (e) the arc length of the circle of radius e running within the surface of the semiconductor wafer around the center of the corresponding rotor disk; the radial position with respect to the center of the working disks; σ _{i is} the angular velocity of the circulation of the carriers around the center of the working disks; ω _i indicates the angular velocity of the spin of the carriers about their respective centers, e _min = max {0; e _exz - R} and e _max = e _exz + R _where R = radius of the wafer denote the lower and upper limits of integration over e; e _{exz indicates} the eccentricity of the wafer in the carrier and the index i = o for the upper working _disk or i = u for the lower working disk indicates whether the angular velocities σ _i and ω _i relate to the upper or the lower working disk.

für jede radiale Position r um weniger als 30% von der über die ganze Arbeitsschicht gemittelten theoretischen Abnutzung abweicht, wobei der Betrag der theoretischen Abnutzung einer jeden Arbeitsschicht gegeben ist durch

wobei die Symbole die beim dritten Verfahren angegebene Bedeutung haben.The object is also achieved according to the invention by a fourth method for simultaneous two-sided grinding of a plurality of semiconductor wafers, wherein each semiconductor wafer is freely movable in a recess of a plurality of offset by means of a rolling device rotors and is thereby moved on a cycloidal trajectory, wherein the semiconductor wafers between two rotating working wheels are machined material, each work disc comprises a working layer containing bonded abrasive, characterized in that the angular velocities σ _i and ω _i are chosen so that the amount of theoretical wear for each working layer

for each radial position r differs by less than 30% from the theoretical wear averaged over the entire work shift, the amount of theoretical wear of each work shift being given by

where the symbols have the meaning given in the third method.

Finally, the object is also achieved by a fifth method for simultaneous double-sided grinding of a plurality of semiconductor wafers (which in turn is not the subject of the invention), wherein each semiconductor wafer is freely movable in a recess of one of a plurality of offset by means of a rolling device rotor discs and thereby on a cycloidal Trajectory is moved, wherein the semiconductor wafers are machined material between two rotating working discs, each working disc comprises a working layer containing bonded abrasive, characterized in that the proportion of the caused by the release of the working layers abrasive material removal on the entire material removal is always less than the proportion of material caused by the fixed in the working layer abrasive abrasive removal.

By means of the aforementioned methods and in particular a meaningful combination of these methods (which is not the subject of the invention), it is possible to produce semiconductor wafers with significantly improved properties.

• – ein isotropes Schliffbild, wobei Bereiche mit Schleifriefen, die parallel oder symmetrisch bezüglich eines Punkts oder einer Symmetrieachse zueinander verlaufen, weniger als 10% der gesamten Oberfläche der Halbleiterscheibe ausmachen,
• – eine Dickenvariation von weniger als 1 μm auf der gesamten Halbleiterscheibe abzüglich eines Randausschlusses von 1 mm,
• – eine auf einen am Rand der Halbleiterscheibe liegenden Bereich mit einer Breite von 1/10 des Durchmessers der Halbleiterscheibe entfallende Dickenvariation von weniger als 0,7 μm,
• – eine auf einen im Zentrum der Halbleiterscheibe liegenden Bereich mit einem Durchmesser von 1/5 des Durchmessers der Halbleiterscheibe entfallende Dickenvariation von weniger als 0,3 μm,
• – einen Warp und einen Bow von jeweils weniger als 15 μm,
• – eine RMS-Rauigkeit von weniger als 70 nm im Korrelationslängen-Bereich von 1 μm bis 80 μm und
• – eine Tiefe der oberflächennahen Kristallschädigung von weniger als 10 μm.

By means of said method, a semiconductor wafer (which is not the subject of the invention) can be produced, which is characterized by

• An isotropic micrograph, wherein regions of scratches which are parallel or symmetrical with respect to a point or axis of symmetry constitute less than 10% of the total surface area of the semiconductor wafer,
• A thickness variation of less than 1 μm on the entire semiconductor wafer minus an edge exclusion of 1 mm,
• A thickness variation of less than 0.7 μm attributable to a region lying at the edge of the semiconductor wafer with a width of 1/10 of the diameter of the semiconductor wafer,
• A thickness variation of less than 0.3 μm attributable to a region in the center of the semiconductor wafer having a diameter of 1/5 of the diameter of the semiconductor wafer,
• A warp and a bow of less than 15 μm each,
• An RMS roughness of less than 70 nm in the correlation length range of 1 μm to 80 μm and
• A depth of near-surface crystal damage of less than 10 μm.

Brief description of the figures

The invention will be further explained with reference to figures.

1 shows a device suitable for carrying out the method according to the invention.

2 shows the lower working disk of the 1 illustrated apparatus with the rolling device, the rotor discs and the semiconductor wafers to be processed in plan view.

3 illustrates the naming and assignment of the characteristic elements for describing the course of motion (kinematics).

4 represents the diametrical thickness profile of a 300 mm diameter single crystal silicon wafer which has been subjected to a grinding process that embodies all the features of the first, second, third, fourth and fifth methods. TTV = 0.62 μm.

5 represents the diametrical thickness profile of a 300 mm diameter single crystal silicon wafer subjected to a grinding process that embodied all the features of the first, second, third and fourth methods. TTV = 1.68 μm.

6 represents the thickness profile of a semiconductor wafer that has been subjected to a grinding process that realized all the features of the second, third, fourth and fifth methods. TTV = 3.9 μm.

7 represents the thickness profile of a semiconductor wafer that has been subjected to a grinding process that realized all the features of the first, third, fourth and fifth methods. TTV = 0.8 μm.

8th shows machine settings (speed sets) and resulting invariant parameter sets (co-rotating reference system). (A): process not carried out according to the invention; (B): Method according to the invention, incorporating the features of the second, third and fourth methods.

9 outputs the parameter sets 8th associated trajectories 19 with respect to upper and 20 with respect to lower working disk again. (A): process not carried out according to the invention; (B): Method according to the invention, incorporating the features of the second, third and fourth methods.

10 shows those from the parameter sets 8th calculated radial wear profiles of the upper 25 and lower 26 working layers. (A): process not carried out according to the invention; (B): Method according to the invention, incorporating the features of the second, third and fourth methods.

11 shows those from the parameter sets 8th calculated differences of the radial wear profiles of the upper and lower working layers. (A): process not carried out according to the invention; (B): Method according to the invention, incorporating the features of the second, third and fourth methods.

12 shows the cumulative and normalized lengths of the machining marks (grinding marks) found on the ground semiconductor wafers as a function of their orientation to Notch (0 °) in the form of a histogram. (A): obtained by the second method; (B): Obtained by a method that does not realize the features of the second method.

List of reference numbers and abbreviations used

• 1 upper working disk
• 4 lower working disk
• 5 Rotary axis of the working wheels
• 7 inner drive ring
• 9 outer drive ring
• 11 upper working shift
• 12 lower working shift
• 13 rotor disc
• 14 Recess in rotor disk for receiving the semiconductor wafer
• 15 Semiconductor wafer
• 16 Center of the semiconductor wafer
• 17 Circle radius of the centers of the carriers in the rolling device
• 18 Point of the semiconductor wafer
• 19 Trajectory of the point of the semiconductor wafer on lower working disk
• 20 Trajectory of the point of the semiconductor wafer on the upper working disk
• 21 Center of the rotor disc
• 22 Center of the rolling device
• 24 Edge region of reduced thickness of the semiconductor wafer
• 25 Wear of the upper working layer
• 26 Wear of the lower working layer
• 27 Area of very high local wear of the work shift
• 28 Range of very high difference of local wear of working shifts
• 29 Difference in wear of upper and lower working shift
• 30 working gap
• 33 Crown of the semiconductor wafer
• 34 Cooling-lubricant bushings
• 35 Isotropic cumulative distribution of processing marks (grinding marks)
• 36 Anisotropic cumulative distribution of the processing marks (grinding marks)
• A.S.A. Wear of the working shift
• a Distance of the center of the rotor disc from the center of the rolling device
• ΔA.S.A. Difference in wear of upper and lower working shift
• e distance of the point of the semiconductor wafer from the center of the rotor disc
• e _exz Distance of the center point of the semiconductor wafer from the center of the carrier (= eccentricity of the wafer in the carrier)
• φ (polar) angle of the point on the semiconductor wafer
• H Local thickness of the semiconductor wafer
• l (e) Length of the circular arc portion of the circular arc around the center of the rotor disk and through the Aufpunkt of the semiconductor wafer, which extends within the surface of a semiconductor wafer
• NCL Normalized cumulative length of machining tracks (per angle class)
• n _o Speed of the upper working disk
• n _u speed of the lower working disk
• n _i speed of the inner rolling device
• n _a speed of the outer rolling device
• r _i circle radius of the inner rolling device
• r _a pitch radius of the outer rolling device
• r radial distance of the point on the semiconductor wafer to the center of the rolling device
• 
  Decrease in the thickness of the working layer due to wear
• R radius of the semiconductor wafer
• RR Radial position on working disk
• ρ Radial position on semiconductor wafer
• s arc length of the trajectory of the point of the semiconductor wafer
• σ angular velocity of rotation of the centers of the carriers around the center of the rolling device ("web speed")
• σ _o web speed with respect to the upper working disk
• σ _u web speed with respect to the lower working disk
• ω angular velocity of the self-rotation of the carriers around their respective centers ("own-rotation speed")
• ω _o own rotational speed with respect to the upper working disk
• ω _u own rotational speed with respect to the lower working disk

Description of the device used

1 shows the essential elements of a device according to the prior art, which is suitable for carrying out the method according to the invention. Shown is the schematic diagram of a two-disc machine for processing disc-shaped workpieces such as semiconductor wafers, as they are for example in DE10007390A1 is disclosed, in perspective view. Such a device has an upper 1 and a lower working disk 4 with collinear axes of rotation 5 and with a substantially plane-parallel arrangement of the working surfaces of the working disks to each other. According to the prior art, the working wheels 1 and 4 Made of cast iron, stainless steel casting, ceramics, composite materials or similar. The work surfaces are uncoated or provided with a coating of, for example, stainless steel or ceramic, etc. The upper working disk contains numerous holes 34 through the work gap 30 Operating resources can be supplied. For the application of such a device as a grinding machine, this is a cooling lubricant (eg water). The device is equipped with a rolling device for runners 13 Mistake. The rolling device consists of an inner 7 and outer drive ring 9 , The runners 13 each have at least one recess, which is a semiconductor wafer to be processed 15 can record. The rolling device can be designed, for example, as a driving stick pin toothing, as involute toothing or as another common toothing type. Preferably, for reasons of ease of maintenance, manufacturing costs and because of the generally large machine dimensions and the associated inevitable play of the transmission elements in this respect uncritical pin toothing. Upper 1 and lower working disk 4 and inner 7 and outer drive rim 9 are at speeds n _o , n _u , n _i and n _a to substantially identical axes 5 driven.

In the case of using the device for a method according to the invention, each working disk carries 1 . 4 on her work surface a working shift 11 . 12 consisting preferably of wipes (woven, knitted, felted, fiber fabric, plastic matrices with or without fiber insert, etc.), foils (single-layer or multi-layer) or foams, in which abrasive material in contact with the semiconductor wafers reaching upper layers abrasive materials are included as an abrasive.

An example of a film suitable for carrying out the process according to the invention is shown in FIG US6007407 disclosed. Examples of wipes are disclosed, for example, in WO99 / 24218 and US5863306 , Examples of such films or cloths with structured (textured, "micro-replicated") work surfaces are shown in FIG US6599177B2 specified.

Preferably, the working layers are glued to the working wheels. According to the prior art such cloths, films or layers are provided on the back with a self-adhesive coating and are fixed by gluing on the working wheels. In particular, in devices with large dimensions, the error-free application of such working layers on the working wheels without errors such as trapped air bubbles, upsetting, stretching or belly of the working layer and also the removal of the working layer after consumption is difficult. So there JP2001-219362A an equipped with pores (channels) execution of such a working layer, can escape through the trapped between the working disk surface and cloth back air bubbles, so as to give a flat, uniform cloth pad. Further suggests WO95 / 19242 An equipment of the cloth back with check mark and a complementary equipped working surface of the working wheels before ("Velcro"), which allow a particularly fast and residue-free change of working shifts. Often the cloths, foils, foams or layers can not be produced in one piece. They are then piece by piece on large carrier documents (foil, cloth, foam, etc.) laminated or assembled. This is disclosed, for example US6179950B1 ,

For carrying out the method according to the invention, moreover, the attachment of the working layers, for example by suction with vacuum (through an air-permeable layer of the working disk made of porous material, such as ceramic), by magnetic or electrostatic attachment or by covering means mounted on the working disk clamping devices, etc. suitable.

The between the on the upper 1 and lower working disk 4 fixed working shifts 11 and 12 formed working gap, within which the semiconductor wafers are processed, is in 1 With 30 designated.

2 shows the device in plan view of the lower working disk 4 , The semiconductor wafers 15 are in runners 13 , which are also referred to as guide cages inserted. The wafers are not firmly connected by form or adhesion with the respective recess of the rotor, so that they can move freely within the recesses. In the preferred case of round semiconductor wafers, in particular a self-rotation of the semiconductor wafers in the recesses of the carrier disks is possible. This self-rotation is desirable because the semiconductor wafers then assume rotationally symmetric shape, which increases their flatness and symmetry and thus advantageous in the context of the invention.

In the following, the center of the working wheels and rolling device, so the entire device, even with 22 designated. The center of a semiconductor wafer 15 in a rotor disc 13 be with 16 , and the center of this rotor disc with 21 designated. An arbitrary Aufpunkt 18 describes on the lower working shift 12 the lower working disk 4 due to the rotation of the working disk and the rotation of the drive rings 7 and 9 a trajectory 19 , The centers 21 the rotor discs 13 walk on a center 22 the rolling concentric pitch circle 17 around.

3 defines further parameters for the description of the movement of the semiconductor wafer in the grinding machine. The reference system is chosen so that it rests in the observed working disk (co-rotating frame of reference). In the supervision in 3 is only the lower working disk 4 located. s denote the arc length of the trajectory 19 of the point of view 18 the semiconductor wafer 15 in a rotor disc 13 about the working shift 12 , The location of this point 18 is described at any time by a radial distance r from the center 22 the rolling device and an angle φ (planar polar coordinates). Because of the turns n _i and n _{a of} the inner 7 and outer drive rim 9 and the rotation of the working disk rotates the rotor disk 13 with angular velocity ω around its center 21 , and this center 21 runs at angular velocity σ around the center 22 the entire device. The distance between the center 21 the rotor disk and the center point 16 the semiconductor wafer 15 is referred to as the eccentricity e _{exz of} the semiconductor wafer in the rotor _disk . Let e be the distance of the point 18 on the semiconductor wafer 15 from the center 21 the rotor disk 13 , R is the radius of the semiconductor wafer 15 , l (e) is the length of the arc with radius e around the center 21 the rotor disk 13 that is inside the surface of the semiconductor wafer 15 runs.

Description of the first method (not the subject of the invention)

According to the first method, the temperature in the working gap is kept constant, preferably throughout the duration of the simultaneous two-sided grinding. For this purpose, the temperature in the working gap is measured during grinding and corrected by suitable measures if the measured temperature deviates from the setpoint temperature. The measurement of the temperature can take place, for example, at fixed intervals or continuously. Due to the constant temperature in the working gap caused by temperature change deformation of the working wheels is avoided and the working wheels are kept in a constant, plane-parallel shape. This results in a significantly improved geometry of the processed semiconductor wafers.

In one embodiment of this first method, each working disk has at least one cooling labyrinth through which a coolant flows. In this embodiment, the temperature or the flow rate of the coolant is suitably varied to counteract an undesirable temperature change and to achieve a constant temperature in the working gap. A suitable and preferred arrangement of cooling labyrinths is in DE19954355A1 disclosed. This arrangement is characterized by an upper layer ("top plate"), which is traversed by a cooling labyrinth, a thermally insulating intermediate layer and a lower layer ("lower plate"), which is traversed by a second cooling labyrinth. Further, there is disclosed a method for adjusting and controlling the planarity of a polishing plate for lapping, grinding or polishing substrate discs, the saucer of an at least three-ply polishing plate is tempered and then the temperature is kept constant and the top plate of the entire working disk is temperature and the temperature the respective polishing process is adapted such that by the temperature control of the saucer stationary heat conditions are created in the polishing apparatus. A corresponding application is also possible with the grinding method according to the invention.

However, it is particularly preferable to keep the temperature in the working gap constant by varying the temperature or the flow rate of the cooling lubricant supplied to the working gap in accordance with the measured temperature. Both parameters, temperature and flow rate, can be suitably varied to keep the temperature in the working gap constant. This kind of Regulation has the advantage over the temperature control of the cooling labyrinths that it is much less sluggish.

If a temperature above the set target value is measured, the temperature of the coolant or the cooling lubricant is lowered in a control loop. If, on the other hand, a temperature below the specified target value is measured, then the temperature of the coolant or of the cooling lubricant is increased, so that the temperature in the working gap remains substantially constant.

The measurement of the temperature in the working gap takes place, for example, directly through temperature sensors built into the surface of the working disks through the (thin) working layer or through small "measuring windows" recessed in the working layer. Since the work wheels rotate during grinding, the temperature reading is transmitted either touching, for example via electrical sliding contacts or non-contact, for example via radio, infrared or inductive. Alternatively, a measurement of the temperature in the working gap can also take place indirectly via a measurement of the temperature of the cooling lubricant flowing out of the working gap.

Description of the second method (not the subject of the invention)

In the following, the second method is described in more detail: In this method, the working disks rotate at higher angular velocity about the center of the entire apparatus than the carriers around their respective centers. More precisely, this means that the angular velocities Ω _i of upper, Ω _o , and lower, Ω _u , working disk are greater in magnitude than the amount of the difference between the angular velocity ω _{0 of} the spin of the carriers about their respective centers and the angular velocity σ ₀ of the rotor Circulation of the rotor discs around the center of the entire rolling device, | Ω _i | ≥ | ω ₀ - σ ₀ |. The spread of the velocity distribution is thereby reduced. The relative speeds between the semiconductor wafers and the working layers of the working disks are not constant due to the process, but dependent on location and time. Speed distribution is the frequency of occurrence of certain relative speeds. A low spread velocity distribution is advantageous because it results in isotropic processing of the semiconductor wafers.

In the context of the second method, the trajectories of the semiconductor wafers relative to each of the two working disks are preferably each epitrochoid, d. H. regular, lengthened or shortened epicycloids.

Furthermore, in the context of the second method, it is preferred that the lengths of the trajectories which cover the semiconductor wafers at the same times relative to the two working wafers are approximately equal. This requirement is considered to be met, in particular, if the amount of the ratio of the difference between the lengths of the trajectories which cover the wafers relative to the two working disks in equal times and the average of the lengths of these trajectories is less than 20%. However, there are also kinematics, which bring a completely equal length of the trajectories, which is not absolutely necessary. Almost equal lengths of the trajectories can be achieved that the rotational speed of the carriers is relatively small compared to the rotational speed of the working wheels.

Due to the aforementioned measures, the front and back sides of the semiconductor wafers experience identical frictional forces at all times, starting directions of the working layers, speeds and accelerations. In particular, abrupt load changes are avoided and supports a uniform rotation of the semiconductor wafers in the holes of the carriers. The speed profiles are similar for both front and back in terms of spread and time distribution. The result is an approximately symmetrical removal of material from the front and rear sides and an isotropic micrograph with low location-dependent or front / rear asymmetric roughness or near-surface damage to the crystal (strain-induced warp / bow). As a result, the surface of the semiconductor wafer becomes plane and isotropic without distortions and deformations, as known, for example, as a "looping nib" (center depression) or "edge drop" (thickness decrease in the edge region) of prior art grinding, lapping or polishing methods. In addition, there is the advantage that the edge profile, which was usually made prior to performing the simultaneous two-sided grinding, is not changed asymmetrically and the symmetry of the edge profile is retained.

Description of the third and fourth inventive method

The third and fourth inventive methods are described in detail below:
Since a working layer with self-sharpening properties is required for carrying out the method according to the invention, the working layer must undergo a certain finite wear in order to constantly expose new, sharp abrasives which lead to a uniform grinding characteristic. On the other hand, too high a wear of the working layer from grinding to grinding is not desirable because then the thickness and shape of the working layer would change too fast and a constant tracking of the processing parameters (machine and process parameters) would be required, resulting in a unfavorable because unstable , Process would lead. So there is an optimal rate of wear, which just guarantees self-sharpening properties, but on the other hand does not lead to a geometrically unstable working layer, so that a largely stable machining process is possible, the reproducible semiconductor wafers with wide ranges constant flatness properties.

In order to be able to predict the wear of the working layer, its load must be determined spatially resolved by the wafers it processes. This requires a precise description of the trajectories the wafers travel over the working wheels during processing.

The trajectory z ⇀ (t) of an arbitrary point can be traced in a frame of reference (invariant frame of reference) moving along with the rotating working disc 18 a semiconductor wafer over a working disk with the in 3 in complex numbers, write z ⇀ (t) = x (t) + iy (t) as z ⇀ (t) = ae ^iσt + ee ^iωt . (1)

With the identity e ^1x = cosx + i · sinx, we obtain immediately from equation (1) the temporal parameter representation of the trajectory in real Cartesian coordinates (x (t); y (t)).

ergeben sich durch Betragbildung und Ableitung nach der Zeit zu r(t) = √a² + e² + 2aecos(σ – ω)t und s .(t) = √a²σ² + e²ω² + 2aeσωcos(σ – ω)t. (2) The radial position r (t) = | z ⇀ (t) | and the amount v (t) ≡ s (t) of the web speed

arise by accumulation and derivative by time r (t) = √ a² + e² + 2aecos (σ - ω) t and s. (t) = √ a²σ² + e²ω² + 2aeσωcos (σ - ω) t , (2)

Where s (t) is the arc length traveled and a point above a variable is its derivative by time.

r(t) aus Gleichung (2) und φ(t) aus Gleichung (3) liefern eine Parameterdarstellung nach der Zeit in ebenen Polarkoordinaten.The angle φ (t) of the position of the point P in plane polar coordinates (r (t); φ (t)) and the time derivative r. (T) of the radial position r (t) are finally given by

r (t) from equation (2) and φ (t) from equation (3) provide a parametric representation of time in plane polar coordinates.

erhält man

Taking into account ∂ / ∂x arctan

you get

Substituting Equation (2) for r (t) into the expressions for s (t), r (t), and φ (t) yields the corresponding terms as a function of the radial position r on the working disk.

der Arbeitsschicht beschrieben werden als proportional zur pro Flächenelement r·∂r·∂φ von Aufpunkt 18 überstrichenen Bogenlänge ∂s und der dazu benötigten Zeit ∂t,

Without further assumptions, this can be achieved by any point of view crossing the working shift 18 a semiconductor wafer 15 caused radial dependent wear

of the working layer are described as being proportional to the per area element r · ∂r · ∂φ of Aufpunkt 18 swept arc length ∂s and the time required ∂t,

By substituting the terms found above, one obtains

Dadurch wird der Beitrag aller gleichwertigen Punkte der Halbleiterscheibe mit identischem Abstand e um den Mittelpunkt der Läuferscheibe, die im Zuge der Eigendrehung der Läuferscheibe alle zu irgendeiner Zeit in gleicher Weise den betrachteten Punkt der Arbeitsfläche überstreichen und zu deren Abnutzung beitragen, berücksichtigt. Integration des erhaltenen Ausdrucks über alle e ergibt schließlich den gesuchten Ausdruck

für die Abnutzung der Arbeitsschicht durch die Gesamtheit aller möglichen Aufpunkte innerhalb der flächig ausgedehnten Halbleitrscheibe,

Finally, the length l (e) of the circular arc with radius e around the center of the rotor disk, which runs through the semiconductor wafer in the rotor disk, is determined numerically for all e in the permitted value range for

As a result, the contribution of all equivalent points of the semiconductor wafer with an identical distance e around the center of the rotor disk, which in the course of the rotor's own rotation all at the same time cover the considered point of the working surface and contribute to its wear, is considered. Integration of the obtained expression over all e finally yields the sought expression

for the wear of the working layer through the totality of all possible Aufpunkte within the flat-extended semiconductor disk,

The index i = o (upper) or i = u (lower working disk) denotes the individual angular velocities σ _o , σ _u , ω _o and ω _u with respect to the respective working disk, and e _min = max {0; e _exz - R} and e _max = e _exz + R. Since the arrangement of the semiconductor wafers in the _{carrier disks} can be carried out in many different ways, there will generally be no analytical expression for l (e) which provides a closed solution for the Integral in equation (8). In practice, therefore, for many values e in the value range {e _min ... E _max }, the value for l (e) is calculated, and instead of the integration in equation (8) a summation is carried out via the integrand over all e. L (e) is sometimes referred to as the "shape function" which describes the placement of the wafers in the carriers.

It now turned out to be advantageous to choose parameter _combinations σ _i and ω _i for given values a and e _{exz of} a device suitable for carrying out the method _according to the invention, in which the wear of the working _layer according to equation (8) is as low as possible over the entire radius of the Working layer varies, which led to the definition of the fourth method according to the invention. This can ensure that the working layers are worn homogeneously, which ensures a durable uniform removal of material from the semiconductor wafers. Irregular ripples in the thickness profile of the ground semiconductor wafers can thus be reliably avoided.

der beiden Arbeitsschichten zum Mittelwert der Beträge der Abnutzung beider Arbeitsschichten für jede radiale Position r der Arbeitsscheiben weniger als 1/1000 beträgt. In diesem Zusammenhang ist es auch bevorzugt, dass die Änderung der Dickenhomogenität der Arbeitsschicht infolge Abnutzung weniger als ein Hundertstel des Betrags der Dickenabnahme der Halbleiterscheiben während der Schleifbearbeitung beträgt, wobei die Dickenhomogenität der Arbeitsschicht als Differenz aus größter und kleinster Dicke über die gesamte in Kontakt mit den Halbleiterscheiben gelangende Fläche der Arbeitsschicht definiert ist.In addition, it is also advantageous if the wear of the working layer according to equation (8) is as similar as possible for the upper and lower working layer, which is reflected in the third method according to the invention. Concretely, the latter means that the amount of the ratio of the difference of the amounts of theoretical wear

the two working shifts to the mean of the amounts of wear of both working layers for each radial position r of the working wheels is less than 1/1000. In this connection, it is also preferable that the change in the thickness homogeneity of the working layer due to wear is less than one-hundredth of the amount of decrease in the thickness of the semiconductor wafers during the grinding, wherein the thickness homogeneity of the working layer as the difference of the largest and smallest thickness is defined over the entire reaching in contact with the semiconductor wafer surface of the working layer.

Preferably, one will select a parameter set for the operation of the grinding device, which meets the requirements of the third and fourth inventive method simultaneously.

erfüllen, erhält man aus den bekannten Gleichungen für Planetengetriebe,

aus den maschinenabhängigen Parametersätzen {n_o, n_u, n_i, n_a} für die Antriebsdrehzahlen n_j(j = o, Drehzahl obere Arbeitsscheibe; j = u, untere Arbeitsscheibe), n_i = Drehzahl des inneren Antriebskranzes und n_a = Drehzahl des äußeren Antriebskranzes und Überprüfung durch Einsetzen in der Formel für

wobei r_i der Teilkreisradius des inneren und r_a der des äußeren Antriebskranzes für die Läuferscheiben sind. Aufgrund der wenigen unabhängigen Freiheitsgrade des Systems liefert dies schnell geeignete Parametersätze, die die Bedingung erfüllen.Suitable machine-independent parameter sets {σ _o , σ _u , ω _o , ω _u }, the condition for

can be obtained from the known equations for planetary gear,

from the machine-dependent parameter sets {n _o , n _u , n _i , n _a } for the drive speeds n _j (j = o, speed upper working disk, j = u, lower working disk), n _i = speed of the inner drive ring and n _a = Speed of the outer drive ring and check by inserting in the formula for

where r _{i is} the pitch circle radius of the inner and r _a of the outer drive ring for the carriers. Due to the few independent degrees of freedom of the system, this quickly provides suitable parameter sets that fulfill the condition.

8 (A) shows an unfavorable combination of parameters {σ _i ; ω _i }, which does not possess these properties and 8 (B) a cheap one that has these qualities. The conversion of the machine-dependent kinematic parameters {n _o , n _u , n _i , n _a } into machine-independent kinematic parameters {σ _o , σ _u , ω _i , ω _a } is explained in more detail in: Th. Ardelt, reports from the technical production Center Berlin, Fraunhofer Institute for Production Systems and Design Technology, IPK Berlin, 2001, ISBN 3-8167-5609-3.

For the in DE10007390A1 disclosed, suitable for carrying out the method according to the invention device with pitch radii r _i and r _{a of} the rolling device for the carriers with ratios r _i / (r _a - r _i ) ≈ 0.552, r _i / (r _a + r _i ) ≈ 0.262 gives the Conversion of the machine-dependent parameter set (n _o , n _u , n _i , n _a ) = (30, -36, -46, 12) RPM the machine-independent parameter set (σ _o , σ _u , ω _o , ω _u ) = (-33.2 , 32.8, 14.0, 80.0) 1 / s.

Focusing on the upper working shift 11 resulting trajectories 19 are in the left half of 9 shown. Focusing on the lower working shift 12 resulting trajectories 20 are in the right half of 9 shown. The working layers indicate the parameter combination 8 (A) an extremely inhomogeneous wear according to equation (8) ( 10 (A) ) on. For the lower working layer is close to the inner edge of a sharply defined area 27 with extremely high local wear and a wider area 25 with some increased wear relative to wear 26 the upper working shift. The difference between the wear of the working layers calculated for these selected process parameters is in 11 (A) shown ( 28 ).

In contrast, shows 8 (B) an inventive choice of process parameters. The obtained wear of the upper and lower working layer ( 25 and 26 ) is symmetrical over the radius of the working disk of the device and almost identical for upper and lower working layer ( 10 (B) ). The difference 29 The wear of both working shifts is over 100 times lower than in the in 8 (A) given example with not inventive parameter selection.

The third and fourth methods of the invention allow the production of semiconductor wafers having the above-described characteristics, the best results being achieved if the requirements of both methods are met simultaneously.

Description of the fifth method (not the subject of the invention)

In the following, the fifth method is described: In this method, the proportion of the material removal caused by the abrasives released during the working layer is always lower than the proportion of material removal caused by the abrasively bonded abrasive in the working layer.

This is achieved in addition to a suitable choice of the average load of the upper working disk in particular and preferably by a uniform load of the working layer over the entire trajectory. For this purpose, it is preferable to keep the temperature in the working gap constant in accordance with the first method in order to avoid deformation of the working disks caused by temperature changes. This results in a working gap between the working layers of the upper and lower working disk that is parallel over the entire process and at each point, and the working layers are loaded with constant force by the semiconductor wafers passed over during processing. Structural collapse of the working layer's grain bonding with premature release of unconsumed cutting grain ("parasitic lapping") due to overload is avoided as well as the equally undesirable exposure of uniform material removal from the wafers due to underload ("cut-in").

The third and fourth methods according to the invention are suitable for achieving uniform loading and thus homogeneous wear of the working layers. Due to the uneven machining forces in non-uniformly worn working layers, the binding of the abrasives contained in the working layers is locally overloaded. The wipes then wear very quickly locally and release excessively unconsumed abrasive. It comes to the so-called "parasitic lapping", d. H. a material removal mainly by free grain as in lapping with lapping slurry. This can be avoided by ensuring uniform wear of the working layers, resulting in semiconductor wafers having significantly less roughness, lower depth of damage, and reduced edge waste.

In addition, this requirement can also be achieved by a homogeneous and little spread speed distribution, which in turn is preferably achieved by the second method. Namely, the material removal rate generally does not necessarily vary in proportion to the pressure and the speed of the grinding movement during grinding due to, for example, a finely-indenting swelling force and due to cooling lubricant and grinding slurry transporting phenomena. Therefore, an inhomogeneous or spread velocity distribution would generally unevenly stress the working layer and result in uneven material removal and thus unwanted resulting shape of the wafer.

Furthermore, it is preferred to choose a sufficient flow rate of the cooling lubricant, which avoids excessive wear of the working layers. Too little coolant results in localized heating of the working layer and thus overloading of abrasive grain (loss of cutting ability), grain bonding or due to thermal expansion and pressure increase of non-uniform wear. Too much cooling lubricant leads to a partial floating of the semiconductor wafers ("Aqua-Planing") and thus also to a deterioration of the uniformity of the material removal.

In particular, it is also preferred that the thickness decrease of the working layer as a result of wear during a grinding process is less than 10%, particularly preferably less than 2%, of the thickness decrease of the semiconductor wafers during the grinding process.

Each of the five methods helps to produce a semiconductor wafer (not part of the invention) having the characteristics described above. However, particularly advantageous properties of the semiconductor wafer are obtained if one meets the requirements of several or, ideally, all methods simultaneously (which is not the subject of the invention).

Preferred embodiments

In the following, preferred embodiments are described which are valid for all methods according to the invention:
As the abrasive bonded in the working layers, a hard material having a Mohs hardness ≥ 6 is preferable. Possible abrasives known in the art include diamond, silicon carbide (SiC), ceria (CeO ₂ ), corundum (alumina, Al ₂ O ₃ ), zirconia (ZrO ₂ ), boron nitride (BN, cubic boron nitride, CBN) Silicon dioxide (SiO ₂ ), boron carbide (B ₄ C) to much softer substances such as barium carbonate (BaCO ₃ ), calcium carbonate (CaCO ₃ ) or magnesium carbonate (MgCO ₃ ). However, particularly preferred are diamond, silicon carbide (SiC) and alumina (Al ₂ O ₃ , corundum).

The mean grain size of the abrasive should be less than 9 μm. The preferred size of the abrasive grains bonded in the working layers is on average 0.1 to 9 μm and more preferably 0.1 to 6 μm in the case of diamond abrasive. The diamonds are preferably incorporated individually or as conglomerates ("clusters") in the binding matrix of the working layer. In the case of a conglomerate bond, the preferred grain diameters are the primary particle size of the cluster constituents.

Working layers with a ceramic bond are preferably used; a synthetic resin bond is particularly preferred; in the case of working layers with conglomerates, also a hybrid bonded system (ceramic bond within the conglomerates and resin bond between conglomerates and working layer matrix).

The hardness of the working layer is preferably at least 80 Shore A. Particularly preferably, the working layer is constructed in multiple layers, wherein the top and bottom have different hardness, so that point elasticity and long-term compliance of the working layer can be adjusted independently of the process requirements.

Before the first use of a working layer, the abrasives incorporated in the working layer are preferably exposed by removing the uppermost layer in order to utilize them for the grinding process. This initial sharpening is carried out, for example, with the aid of grindstones or knives, which are preferably mounted on specially modified carriers and, similar to the method according to the invention, are guided over the two working disks by means of the rolling device. Initial sharpening is also called "dressing" in English.

Preferably, the dressing is done with grindstones containing abrasive grain having a similar grain size to the abrasive in the working layers. These "sharpening stones" may, for example, be inserted annularly and into an externally toothed driving ring, so that they can be guided by means of the rolling devices of the grinding machine in a suitable manner between the upper and lower working layers. The sharpening stones preferably coat the entire surface of the working layers during the dressing, and particularly preferably run even temporarily or constantly beyond their edge. Preferably, the abrasive grain is bonded in the sharpening stone that the wear of the sharpening still allows economic sharpening operation, but during the sharpening always at least one layer of loose sharpening stone grain is in the working zone between sharpening stone and working surface, so that sharpening predominantly done by free (unbound) grain.

It has been shown that the sharpening process produces a disturbed near-surface layer in the working layer whose depth has approximately the extent of the sharpening grain. A sharpening stone with too coarse grain therefore imposes a structure on the working layer, which is characterized by the grain of the sharpening stone and not by the properties of the working layer. This is unfavorable for the desired uniform self-sharpening of the working layer in the subsequent grinding operation. Too fine sharpening stone provides too little material removal and leads to an uneconomical sharpening process. Finally, it has been found that sharpening predominantly by free sharpening grain as a result of the rolling motion of the sharpening grain during the sharpening movement exerts less directed forces on the working layer than sharpening with predominantly solid sharpening grain and although a rougher, but particularly isotropic sharpened working layer results.

Preferably, a grain that is softer than the abrasive grain used in the working layer is used for sharpening or dressing the working layer. Particularly preferred is the sharpening grain of corundum (Al ₂ O ₃ ).

In the operation of the invention according to the claimed method with a suitable choice of working layer and machine parameters by continuous wear of the working layer dulled abrasive residues are removed and constantly exposed new, cut-friendly abrasives. As a result, a continuous operation until the wear of the working layers is completely possible. This operating condition without intermediate re-sharpening operation is referred to as "self-sharpening working" of the working layers and is particularly preferred. The intervention of the exposed on the surface of the working layers grains in the surface of the semiconductor wafers and taking place by the relative movement of the working layer and semiconductor wafers material removal is technically referred to as "multi-grain loops with geometrically undefined cutting edge".

Preferably, the grinding is operated so that the selected speeds of the drives of the grinding device lead to flat as possible semiconductor wafers. Due to the kinematic coupling of tool and workpiece movement ("planetary gear"), the movement of the working disks can then no longer be selected independently. In particular, motion sequences may occur in which the wear of the working layers no longer takes place completely homogeneously over their entire surface. Thus, the work shifts slowly lose their initial shape, and it may be necessary to occasionally dress the work shifts to restore a plane-parallel working gap. Preferably, the working layer is chosen so that self-sharpening operation is achieved with the least possible wear, and the drives set so that even the best possible shape of the wafer as uniform as possible loading of the working layer takes place, so that such interim dressing operations must be as rare as possible. For a desired TTV of the semiconductor wafer of less than 1 μm, operation is still regarded as economic, if at most every 20th journey requires training; for a TTV smaller than 2, if at most after every 50th journey must be trained.

Furthermore, it is preferred that the material removal is effected by predominantly surface engagement of the working layer. By "surface engagement" is meant that the portion of the surface of the working layer that is actually in contact with the wafer during grinding is significantly larger than the contact surface of the abrasive coating of a cup grinding wheel when processed by a conventional cup grinding wheel grinding process , for example, DDG or SSG. (In the case of the DDG, the contact surface of the abrasive coating of the cup grinding wheel makes about 0.5% to 3% of the surface of the semiconductor wafer, in SSG this is about 0.5% to 5%.) Preferably, in the process according to the invention the proportion is greater than 5 % and more preferably 10% to 80%.

It is also preferable that the parts of the carriers that come into contact with the working layers do not contain any metal. The carriers are preferably made of a completely metal-free material, for example a ceramic material. However, it is also preferred rotor wheels with a core of steel or stainless steel, for example, which are coated with a non-metallic coating. Such a coating is preferably made of thermoplastic plastics, ceramics, or organic-inorganic hybrid polymers, such as Ormocer ^® (a silicate compound), diamond ( "diamond-like carbon", DLC), but alternatively also consist of a hard chrome plating or nickel-phosphorus coating.

In the case of carriers made of metal or with a metal core, the walls of the recesses for receiving the semiconductor wafers are preferably lined with a ceramic material, so that no direct contact between the semiconductor wafer and the metal of the rotor disc is formed.

Preferably, the recesses for receiving the semiconductor wafers are mounted in the carriers such eccentric with respect to the center of the respective rotor, that the center of the rotor is outside the surface of the semiconductor wafers. For example, when processing semiconductor wafers having a diameter of 300 mm, this is an eccentricity of more than 150 mm relative to the center of the carrier. A rotor disk preferably has three to eight recesses for semiconductor wafers. During a grinding operation, preferably, five to nine carriers are simultaneously in the grinding machine.

mit der sich beliebige Aufpunkte 18 der Halbleiterscheiben 15 über die Arbeitsscheiben 1 und 4 bewegen, ist ein Bereich von 0,02 bis 100 m/s bevorzugt und ein Bereich von 0,02 bis 10 m/s besonders bevorzugt. Aufgrund der Einschränkungen, die beispielsweise die in DE10007390A1 beschriebene geeignete Vorrichtung bezüglich der realisierbaren Drehzahlen der Hauptantriebe aufweist und die typisch für nach dem Stand der Technik verfügbare, zur Durchführung des erfindungsgemäßen Verfahrens geeignete Vorrichtungen ist, ist ein Bereich von 0,2 bis 6 m/s für die Bahngeschwindigkeit besonders bevorzugt.For the amount s (t) = v (t) = | v ⇀ (t) | the web speed

with which arbitrary Aufpunkte 18 of the semiconductor wafers 15 over the work disks 1 and 4 a range of 0.02 to 100 m / s is preferable, and a range of 0.02 to 10 m / s is particularly preferable. Due to the limitations of, for example, the in DE10007390A1 A suitable range of speeds from 0.2 to 6 m / s for the web speed is particularly preferred, as described for suitable devices known in the art and suitable for carrying out the method of the invention.

The pressure with which the working layers are pressed against the semiconductor wafers during processing and the web speed of the wafers over the working layers are preferably selected during the main load step such that the total removal rate, ie the sum of the removal rates on both sides of the semiconductor wafers is 2 to 60 μm / min. The main load step is to be understood as meaning the processing phase within which the greatest proportion of the total removal of the entire grinding treatment is effected, whereby the processing phase is again a period of time during which all process parameters remain constant. As a rule, the main load step is the processing phase with the highest pressure or the proportionately longest duration or both. In the case of a working shift with Diamond abrasive grains with a mean size of 3 to 15 μm, a removal rate between 2.5 and 25 μm / min is particularly preferred.

For the pressure exerted by the working wheels on the semiconductor wafers during the main load step, a range of 0.007 to 0.5 bar is preferred, and a range of 0.012 to 0.3 bar is particularly preferred. For this indication, the pressure is based on the total area of the semiconductor wafers for processing in the device and not on the effective contact area between working layer and semiconductor wafers.

Furthermore, an opposite rotation of the working wheels with respect to the average rotational speed of the carriers during the main load step of processing is preferred.

In addition, it is particularly preferred that the prints, speeds and thus web speeds assume different values for the different processing phases. Finally, it is also particularly preferred that in certain low-pressure processing phases ("Ausfeuer" or "spark out" phases) the working disks rotate in the same direction. Such a Ausfeuer phase is particularly useful at the very end of the entire grinding treatment and therefore preferred.

The cooling lubricant used in the process of the invention preferably consists of a water-based mixture of one or more of the substances mentioned below: Viscosity-modifying additives, in particular viscosity-increasing additives such as glycols, for. Example, short or longer chain polyethylene glycols, alcohols, sols or gels (eg., Additions of fumed silica) and similar substances that are known as coolants or lubricants. Further, pH modifying additives such as acids, bases, and composite buffer solutions are preferred. Particularly preferred are alkaline additives such as potassium hydroxide (KOH), potassium carbonate (K ₂ CO ₃ ), tetramethylammonium hydroxide (N (CH ₃ ) ₄ OH), tetramethylammonium carbonate (N (CH ₃ ) ₄ CO ₃ ), ammonium hydroxide (NH ₄ OH) and sodium hydroxide (NaOH). The pH of the cooling lubricant is preferably in the range 7.0 to 12.5. Furthermore, complexing agents can be added, in particular those which form copper complexes. A particularly preferred coolant but is also pure water without any addition.

The amounts of cooling lubricant supplied to the working gap via the passage in the upper working disk are preferably in the range between 0.2 and 50 l / min and more preferably between 0.5 and 20 l / min. The values given are average values measured by means of a complete grinding treatment and relate to approximately 1.5 m ^{2 of} effective working disk surface, such as, for example, those suitable for carrying out the method according to the invention DE10007390A1 has disclosed device.

The method according to the invention is preferably used for processing semiconductor wafers of monocrystalline silicon with a diameter of greater than or equal to 100 mm, particularly preferably with a diameter of 300 mm or greater. The preferred initial thickness before processing with the method according to the invention is 500 to 1000 microns. For silicon wafers having a diameter of 300 mm, an initial thickness of 775 to 950 μm is particularly preferred.

The semiconductor wafers are processed after the separation of the semiconductor rod in slices (for example by means of a wire saw, band saw or Innenlochsäge) and before the final finish processing (for example by means of a chemo-mechanical polishing) by the method according to the invention. Further processing steps between separating and the method according to the invention or between the method according to the invention and the final finish processing can optionally be added without impairing the suitability of the claimed features of the inventive method for solving the underlying object. This may be, for example, further mechanical, chemical or chemo-mechanical processing steps from groups b), c) and d) of the processing sequence for the production of semiconductor wafers, as indicated in the prior art (see above).

The final thickness of the semiconductor wafers after processing by the method according to the invention is preferably 500 to 950 μm and more preferably 775 to 870 μm. The total removal, d. H. the sum of the individual abrasions from both sides of the semiconductor wafer is preferably 7.5 to 120 μm and more preferably 15 to 90 μm.

It is preferred to precede the grinding process of the invention by cutting a mechanical machining process according to the prior art after the semiconductor rod has been cut into slices.

It is further preferred to allow the finishing process according to the invention to follow further finishing processes according to the prior art before the final finish machining.

Finally, it is preferable to supplement the grinding method according to the invention between the bar cutting and the finish machining by pre- and post-processing steps by methods according to the prior art.

It is particularly preferred to subject the semiconductor wafers directly after the separation of the rod to the grinding method according to the invention and subsequently to a chemo-mechanical polishing and moreover to carry out any further material-removing processing steps. In particular etching processes, lapping treatments or grinding treatments are to be understood as material, in which the material thickness removed by the semiconductor wafers is greater than the thickness variation (TTV) remaining on the semiconductor wafers according to the method of the invention. In this sense, as non-material-removing steps such as cleaning, etching, grinding or polishing steps with material removal of less than on the inventively processed semiconductor wafers remaining thickness variation (TTV), or even measuring, sorting and not significantly the surface of Semiconductor wafer-changing steps such as edge rounding or polishing should not be excluded.

Description of the semiconductor wafer (not the subject of the invention)

Result of the application of the described method, in particular a suitable combination of some or preferably all methods, is a semiconductor wafer with a small thickness variation, the residual unevenness not by a so-called "loop" (local thickness decrease in the disk center) or a so-called "edge drop" (thickness decrease in the edge region of the semiconductor wafer) is decisively determined and whose surface has a largely isotropic, in particular not centro- or radially symmetric distribution of the processing marks designated as grinding marks and a roughness of less than 70 nm RMS.

• – Ein isotropes Schliffbild, wobei Bereiche mit Schleifriefen, die parallel oder symmetrisch bezüglich eines Punkts oder einer Symmetrieachse zueinander verlaufen, weniger als 10% der gesamten Oberfläche der Halbleiterscheibe ausmachen. Die Bestimmung des Grades der Isotropie des Schliffbildes wird nachfolgend erläutert.

The semiconductor wafer in particular has the following advantageous properties:

• An isotropic micrograph wherein areas of scorelines that are parallel or symmetrical with respect to a point or axis of symmetry make up less than 10% of the total surface area of the semiconductor wafer. The determination of the degree of isotropy of the micrograph will be explained below.

• – Eine Dickenvariation von weniger als 1 μm auf der gesamten Halbleiterscheibe abzüglich eines Randausschlusses von 1 mm, wobei eine Dickenvariation von bis zu 50 nm oder sogar weniger erreichbar ist. Der Begriff „Dickenvariation” ist im Sinne des gebräuchlichen Parameters „TTV” (total thickness variation) zu verstehen.
• – Eine auf einen am Rand der Halbleiterscheibe liegenden Bereich mit einer Breite von 1/10 des Durchmessers der Halbleiterscheibe entfallende Dickenvariation von weniger als 0,7 μm, wobei auch Werte von 50 nm oder weniger erreichbar sind. Die Halbleiterscheibe weist somit keinen nennenswerten Randabfall auf.
• – Eine auf einen im Zentrum der Halbleiterscheibe liegenden Bereich mit einem Durchmesser von 1/5 des Durchmessers der Halbleiterscheibe entfallende Dickenvariation von weniger als 0,3 μm, wobei auch Werte von 50 nm oder weniger erreichbar sind. Die Halbleiterscheibe weist somit keinen nennenswerten Schleifnabel auf.
• – Einen Warp und einen Bow von jeweils weniger als 15 μm, wobei auch Werte von 1 μm oder weniger erreichbar sind. Der Parameter „Warp” ist gemäß ASTM F 1390 und DIN 50441-5 definiert, der Parameter „Bow” gemäß ASTM F 534 and DIN 50441-5.
• – Eine RNS-Rauigkeit von weniger als 70 nm, wobei auch Werte von 1 nm oder weniger erreichbar sind. Die angegebenen Werte beziehen sich auf einen Korrelationslängen-Bereich von 1 μm bis 80 μm.
• – Eine Tiefe der oberflächennahen Kristallschädigung von weniger als 10 μm und bis hin zu 0,2 μm und weniger.

12 shows the accumulated (cumulative) lengths of the grinding marks on a semiconductor wafer per angle class as a measure of the isotropy of the processed semiconductor wafer (histogram in plane polar coordinates). The added lengths are indicated normalized to the average grinding length over all angles. 12 (A) has the largely evenly distributed, in total largely the same length processing marks 35 a semiconductor wafer with an isotropic microsection (variation of the cumulative grinding distance lengths per angle class less than ± 10% compared to the mean accumulated grinding radius length over all angles). 12 (B) gives the sanding histogram 36 anisotropic semiconductor wafer again. To determine the values, the surface of a semiconductor wafer is visually inspected and the number attributable to each angle class (here: every 15 °, within ± 7.5 °) multiplied by the length of the grinding marks. Since in grinding processes the size and depth of the scorings are similar to the dimensions of the abrasive grit used, such a method is largely reliable and practicable, largely without ambiguity, by contributions of very fine or very coarse scores within the given limits (± 10%). Alternatively, for example, a less complex angle-resolved scattered light method can be used in which the gloss of the semiconductor surface (extra-specularly scattered light) is measured angle-dependent and whose angular variation is used as a measure of the isotropy of the surface. The angles are given relative to the notch of the semiconductor wafer (Notch = 0 °).

• - A thickness variation of less than 1 micron on the entire semiconductor wafer minus an edge exclusion of 1 mm, with a thickness variation of up to 50 nm or even less achievable. The term "thickness variation" is to be understood in the sense of the common parameter "TTV" (total thickness variation).
• A thickness variation of less than 0.7 μm attributable to a region lying at the edge of the semiconductor wafer with a width of 1/10 of the diameter of the semiconductor wafer, whereby values of 50 nm or less can also be achieved. The semiconductor wafer thus has no appreciable edge waste.
• A thickness variation of less than 0.3 μm attributable to a region in the center of the semiconductor wafer with a diameter of 1/5 of the diameter of the semiconductor wafer, whereby values of 50 nm or less can also be achieved. The semiconductor wafer thus has no appreciable loop.
• - A warp and a bow of less than 15 microns, with values of 1 micron or less can be achieved. The parameter "Warp" is defined according to ASTM F 1390 and DIN 50441-5, the parameter "Bow" according to ASTM F 534 and DIN 50441-5.
• An RNA roughness of less than 70 nm, whereby values of 1 nm or less can also be achieved. The values given refer to a correlation length range of 1 μm to 80 μm.
• - A depth of near-surface crystal damage of less than 10 microns and down to 0.2 microns and less.

Examples

For the conclusion of Examples 1 to 4 described below 4 to 7 a device was used whose features relevant to the invention in DE10007389A1 are described and have already been explained above (Polisher Peter Wolters AC-1500P3). For the examples described below different "Trizact ^® Diamond Tile" were glass-grinding cloths used as work shifts, which were provided by the company. 3M, USA, and is available for example in US6007407 are described. The wipes are self-adhesive on the back and glued to the working disks of the double-side processing device. The cloths used in the following examples were filled with diamond abrasive. The particle size distribution was 2-6 microns. For the wipes used in Examples 1, 3 and 4, the abrasive was firmly bound; in Example 2, however, only loose so that the abrasive pad wore quickly and functioned as a "free-grain" dispenser in engagement with the workpieces.

As workpieces, 300 mm single-crystal silicon disks having output surface as obtained after cutting (wire sawing) were used. They had a starting thickness of 915 μm. The material removal was in all examples 90 microns, the final thickness after machining so 825 microns. The semiconductor wafers were placed in carriers of glass-fiber reinforced epoxy resin (EP-GFRP), which had an initial thickness of 800 microns (thickness decrease due to wear), The feed consisted of five carriers, each with a semiconductor wafer. The pressure of the working wheels during machining on the workpieces was around 340 daN and was increased or decreased so that an average removal rates of 10-20 μm / min were achieved.

Water was used as cooling lubricant (deionized ultrapure water), which was fed through the holes in the upper working disk to the working gap at a rate of between 3 and 20 l / min.

example 1

4 shows the thickness profile of a semiconductor wafer of monocrystalline silicon of 300 mm diameter, which was obtained by processing with a method according to the invention with all features of the first, second, third, fourth and fifth method. The thickness profile was determined by averaging 4 diametrically running individual measurements at 0 °, 45 °, 90 ° and 135 ° relative to the orientation notch of the semiconductor wafer. The thickness variation over the entire semiconductor wafer (total thickness variation, TTV) is determined taking into account all measured thickness values and is 0.62 μm in this example. The thickness profiles were determined by means of a capacitive measuring method, in which a pair of opposing measuring probes determines the distances to the front and back of the semiconductor wafer guided between them. The edge exclusion (non-measurable edge area of the semiconductor wafer) is 1 mm. H in the diagram denotes the thickness of the semiconductor wafer (in microns), ρ the radial position of the respective thickness measurement value (in millimeters).

Example 2

5 shows the thickness profile of a semiconductor wafer of a semiconductor wafer not processed according to the invention. The removal of material from the semiconductor wafer was mainly caused by free (unbound) grain during processing ("parasitic lapping"). Due to the need for full surface material removal transport of the free grain from the free working gap over the edge of the wafer to the center and as a result of the loss of Schnittfrudigkeit the grain on this way (wear) occurs depletion of abtragsfähigem grain from the edge to the center of the semiconductor wafer. The material removal is therefore higher at the edge than in the center of the semiconductor wafer. The result is a convex shape of the semiconductor wafer with the edge decreasing thickness ("edge drop") 24. The TTV is 1.68 microns.

Example 3

6 shows the thickness profile of a semiconductor wafer after machining with a device suitable for carrying out the claimed method according to the invention, but with deformed working disks.

Since the working disks are composed of different materials with correspondingly different coefficients of thermal expansion, a certain unavoidable deformation always occurs in the case of an unsuitable choice of the temperature due to the "bimetallic effect". Furthermore, such a disturbance of the plane parallelism by time-dependent temperature entry during the machining process itself, for example, by the working gap 30 done chip work (which leads to heating) take place; because this results from the processing zone 30 in the work disks 1 and 4 into a temperature gradient, which deforms the working disks (time-dependent). The so processed wafers have a pronounced crown 33 on (high thickness in the center and small thickness in the edge area).

In the in 6 In the example shown, during the processing, insufficient measures were taken to keep the temperature in the working gap constant (unsuitable choice of the double cooling system temperatures of the working disk, insufficient control of the temperature and amount of cooling lubricant supplied to the working gap.) The TTV in this example obtained semiconductor wafer is 3.9 microns.

Example 4

7 shows the thickness profile of a semiconductor wafer after processing in a device, with according to the invention uniform wear of the working layer (shape constancy) and kept constant temperature and working disk shape, but with a kinematics that did not meet the requirements of the second method. The amount of the difference between the rotational speed of the carriers and the rotational speed of the carriers around the center of the rolling device was somewhat larger than the amount of rotational speed of the carriers relative to the working disks, so that the wafers described epitrochoid with respect to one working disk and hypotrochoid with respect to the other working disk. Although the drive speeds selected in the example were outside, but still close to the range according to the second method, results in a rather good TTV of 0.8 microns.

Claims (2)

A method for simultaneous double-sided grinding of a plurality of semiconductor wafers, wherein each semiconductor wafer is freely movable in a recess of one of a plurality of carriers rotated by a rolling device and thereby moved on a cycloidal trajectory, wherein the semiconductor wafers are machined between two rotating working wheels material, each Working disk comprises a working layer containing bonded abrasive, characterized in that the angular velocities σ _i and ω _i are chosen so that the amount of the ratio of the difference of the amounts of theoretical wear

of the two working layers to the mean of the amounts of wear of both working layers for each radial position r is less than 1/1000, wherein the amount of theoretical wear of each working layer is given by

where a is the pitch radius of the orbital movement of the carriers on the working wheels around the center of the rolling device; e is the distance of the currently considered point from the center of the corresponding rotor; l (e) the arc length of the circle of radius e running within the surface of the semiconductor wafer around the center of the corresponding rotor disk; the radial position with respect to the center of the working disks; σ _{i is} the angular velocity of the circulation of the carriers around the center of the working disks; ω _i indicates the angular velocity of the spin of the carriers about their respective centers, e _min = max {0; e _exz - R} and e _max = e _exz + R _where R = radius of the wafer denote the lower and upper limits of integration over e; e _exz the eccentricity of the semiconductor wafer in the rotor disk and the index i = o for the upper working disk or i = u for the lower working disk indicates whether the angular velocities σ _i and ω _i relate to the upper or the lower working disk. A method for simultaneous double-sided grinding of a plurality of semiconductor wafers, wherein each semiconductor wafer is freely movable in a recess of one of a plurality of carriers rotated by a rolling device and thereby moved on a cycloidal trajectory, wherein the semiconductor wafers are machined between two rotating working wheels material, each Working disk comprises a working layer containing bonded abrasive, characterized in that the angular velocities σ _i and ω _i are chosen so that the amount of theoretical wear for each working layer

for each radial position r differs by less than 30% from the theoretical wear averaged over the entire work shift, the amount of theoretical wear of each work shift being given by

where a is the pitch radius of the orbital movement of the carriers on the working wheels around the center of the rolling device; e is the distance of the currently considered point from the center of the corresponding rotor; l (e) the arc length of the circle of radius e running within the surface of the semiconductor wafer around the center of the corresponding rotor disk; the radial position with respect to the center of the working disks; σ _{i is} the angular velocity of the circulation of the carriers around the center of the working disks; ω _i indicates the angular velocity of the spin of the carriers about their respective centers, e _min = max {0; e _exz - R) and e _max = e _exz + R _where R = radius of the wafer denote the lower and upper limits of integration over e; e _{exz indicates} the eccentricity of the wafer in the carrier and the index i = o for the upper working _disk or i = u for the lower working disk indicates whether the angular velocities σ _i and ω _i relate to the upper or the lower working disk.

Images