Method for producing a monocrystalline silicon rod, and monocrystalline silicon rod
The method addresses the inefficiencies of the Czochralski process by controlling crucible rotation and heater power to produce a dislocation-free silicon rod without a large end cone, reducing material waste and processing time.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- SILTRONIC AG
- Filing Date
- 2025-12-05
- Publication Date
- 2026-06-18
AI Technical Summary
Existing methods for producing single-crystal silicon rods in the Czochralski process result in significant material waste and prolonged processing time due to the need for a large end cone to ensure dislocation-free separation, which is difficult to measure and control accurately.
A method involving controlled crucible rotation, reduced drawing speed, increased heater power, and a CUSP magnetic field to produce a dislocation-free silicon rod without an explicit end cone, allowing for a significantly shorter end piece and precise diameter control using optical sensors.
Reduces material waste and processing time by producing a dislocation-free silicon rod with a very short end piece, ensuring efficient use of high-purity silicon and minimizing machine downtime.
Smart Images
Figure EP2025085660_18062026_PF_FP_ABST
Abstract
Description
[0001] 202400024 / Kl
[0002] Method for producing a monocrystalline rod made of silicon and monocrystalline rod made of silicon
[0003] The invention relates to a method for producing a monocrystalline silicon rod, comprising melting polysilicon in a crucible, drawing a single crystal containing an initial cone, a cylindrical section, and an end cone. The invention also relates to a single-crystal silicon rod.
[0004] Single-crystal silicon, the starting material for most processes for manufacturing electronic semiconductor devices or photovoltaic modules, is usually produced using the so-called Czochralski process ("CZ"). In this process, polycrystalline silicon ("polysilicon") is placed in a crucible and melted, a seed crystal is brought into contact with the molten silicon, and a single crystal is grown by slow extraction.
[0005] The crucible is typically made of a silicon dioxide-containing material such as quartz. It is usually filled with fragments and / or granules of polycrystalline silicon, which is melted using a side heater arranged around the crucible and a bottom heater located underneath. After a period of thermal stabilization of the melt, a single-crystal seed crystal is immersed in the melt and lifted. Silicon then crystallizes at the end of the seed crystal wetted by the melt. The rate of crystallization is primarily influenced by the rate at which the seed crystal is lifted (crystal lift rate) and by the temperature and temperature gradient at the interface where the molten silicon crystallizes.By appropriately controlling these parameters, a section called the thin neck is first extracted to eliminate dislocations, then a conical section of the single crystal (initial cone) and finally a cylindrical section of the single crystal, from which the semiconductor wafers are later separated.
[0006] When growing single crystals from semiconductor material using the Czochralski method, the diameter of the single crystal must be monitored, i.e., determined and, if necessary, influenced. The change in the diameter of the single crystal, which depends on the radial growth of the single crystal at the crystallization boundary, can be [202400024 / Kl]
[0007] 2 can be influenced, for example, by selectively changing the lifting speed of the single crystal and / or the temperature of the melt in the area of the crystallization limit.
[0008] The diameter of a single crystal can typically be determined using optical instruments such as a camera. This is done by capturing, for example, three points on a bright ring surrounding the single crystal in the region of the crystallization boundary and calculating the diameter from these points. This bright ring is a reflection from the glowing wall of the crucible containing the melt, which forms a meniscus in the region of the crystallization boundary.
[0009] Towards the end of the drawing process of a cylindrical part, a so-called end cone is typically drawn on the single crystal; that is, after a long cylindrical section, the single crystal tapers conically to a small diameter. This prevents any glide or dislocations that occur during the separation of the single crystal from the melt from propagating far back into the single crystal, particularly into the cylindrical part, which is subsequently used, for example, in the production of wafers.
[0010] Since the camera mentioned is usually located outside the associated device for pulling the single crystal due to the high temperatures in the area of the melt, and therefore usually only looks down at the melt at a relatively steep angle, the diameter of the single crystal in the area of the end cone can usually no longer be measured.
[0011] For example, a very long end cone can be drawn at an angle at which the camera can just barely capture the bright ring. However, this consumes a very large amount of material that cannot be used afterward and may have to be remelted. Therefore, the larger the diameter of the single crystal in the cylindrical region, the greater the amount of material that is unnecessarily wasted in the end cone.
[0012] From EP 0 758 690 A1, for example, a method is known in which the diameter of the single crystal in the region of the end cone is detected by means of a camera 202400024 / Kl
[0013] Method 3 involves positioning a mirror in an area above the surface of the melt. The mirror allows the camera to see the bright ring even at smaller angles. However, a disadvantage of this method is that such a mirror is difficult to install and also fogs up very easily due to the vapors rising from the melt.
[0014] From JP 63 021 280 A1, a method is known in which the position of the camera is changed during the drawing of the end cone in order to capture a less steep angle. However, this is disadvantageous because, firstly, changing the camera's position requires additional effort, and secondly, a precisely positioned camera is necessary for a reproducible single crystal shape, which cannot be reliably guaranteed if the camera's position is constantly changed.
[0015] From EP 0 498 653 A2, a method is known in which the sinking of the melt surface is determined from the weight of the single crystal pulled from the melt. From this sinking, for example, a height by which the crucible must be advanced can be determined, or an optical diameter determination of the single crystal can be corrected, for example by correcting the value of the camera height above the melt.
[0016] From DE 42 31 162 A1, a method is known in which the height of the surface of the melt in the crucible is controlled. For this purpose, a distance between the surface of the melt and a reference mark is determined.
[0017] For example, US patent 6 106 612 A discloses a method in which the position of the surface of a melt in a crucible from which a single crystal is pulled is determined relative to a fixed point.
[0018] From EP 3 411 515 A1, a method is known by which the length of the end cone can be shortened by evaluating the settling velocity of the melt surface relative to the crucible and using it to control the diameter in the end cone. 202400024 / Kl
[0019] 4
[0020] While all the aforementioned methods of the prior art offer improvements, there is still a need to make the end cone even smaller in order to save more material, usually high-purity silicon, and simultaneously reduce machine time. The object of this patent is therefore to provide a process that makes it possible to produce a very short end cone that is also dislocation-free. Another object of this patent is to provide a crystal that fulfills the necessary properties.
[0021] This task is solved by the methods and products described in the claims.
[0022] 202400024 / Kl
[0023] 5
[0024] Brief description of the characters
[0025] Figure 1 shows a typical setup for Czochralski crystal growth. Essential components of this setup are a supported crucible 2, resting on a rotatable and liftable crucible shaft 1, containing a silicon melt 3. Around the crucible are a resistance heater 4, designed as a side heater for melting polycrystalline silicon and supplying heat to the melt from the side, and three magnetic coils 5, 6, 7 of a magnetic device for generating a magnetic field that influences melt convection. The application of a CUSP magnetic field is based on the fact that two coils (for example, coils 5 and 7) are positioned opposite each other and each current flows through them in opposite directions, so that each magnetic field is directed towards the other coil.
[0026] A resistance heater 8 is arranged beneath the crucible, acting as a bottom heater to supply heat to the melt from below. Another heating element 9 is located just above the melt 3, acting as an annular heater to reduce the axial temperature gradient G at the boundary between the single crystal 10 pulled from the melt and the melt 3. A heat-insulating heat shield 11 and a cooling jacket 12 are arranged around the single crystal. A special component of the configuration is a molybdenum gas guide tube 13, which is attached to the heat shield and widened at its lower end to form a ring 14.
[0027] Figure 2 shows the end cone manufactured according to the prior art. At the end of a substantially cylindrical section with a diameter D, the diameter tapers continuously over a length L until the rod ends. Lines are shown as examples to represent the shape of the striation S.
[0028] Figure 3 shows an inventive end piece of a rod produced according to a method according to the invention. The rod, with a diameter D, tapers very rapidly over a length L, forming a convex surface with a small elevation in the center. The length L of the end piece is measured from the point where the rod tapers to the tip of the resulting elevation. Here, too, lines are shown as examples to represent the shape of the resulting striations. Near the convex surface, the striations run parallel to this surface. 202400024 / Kl
[0029] 6
[0030] Figure 4 shows an alternative inventive end piece of a rod produced according to a modified method of the invention. The rod, with a diameter D, tapers first slowly to form a truncated cone and then very rapidly to form a convex surface with a small elevation in the center. The length L of the end piece is measured from the point where the rod tapers to the tip of the resulting elevation. Here, too, lines are shown as examples to represent the shape of the resulting striations. Near the convex surface, the striations run parallel to this surface.
[0031] 202400024 / Kl
[0032] 7
[0033] Detailed description of embodiments according to the invention
[0034] For the pulling of silicon single crystals with a nominal diameter of up to 300 mm according to the inventive method, a device was used whose heat-transfer-determining configuration (“hot zone”) is schematically shown in Fig. 1. Elementary components of this configuration are a supported crucible 2 resting on a rotatable and liftable crucible shaft 1, containing a silicon melt 3.
[0035] Around the crucible are a resistance heater 4, acting as a side heater for melting polycrystalline silicon and supplying heat to the melt from the side, and a coil arrangement of a magnetic device for generating a CUSP magnetic field that influences melt convection. Below the crucible is a resistance heater 8, designed as a bottom heater for supplying heat to the melt from below.
[0036] Another heating element 9 is located just above the melt 3, designed as a ring heater to reduce the axial temperature gradient G in the boundary region of the phase boundary between the single crystal 10 pulled from the melt and the melt 3.
[0037] A heat-insulating heat shield 11 and a cooling jacket 12 are arranged around the single crystal. Another component of the configuration is a molybdenum gas guide tube 13, which is attached to the heat shield and widens at its lower end to form a ring 14.
[0038] Furthermore, the housing of the crystal growing device includes a window in front of which optical sensors designed as cameras are mounted. The camera has a detection area that is directed between the single crystal and the heat shield towards the surface of the melt. This optical sensor is used to determine the current diameter of the crystal.
[0039] To perform the Czochralski single crystal growing method, a crucible filled with polycrystalline silicon is heated until the silicon melts. A seed crystal is then seeded after melting using the 202400024 / Kl
[0040] 8
[0041] The surface of the melt is brought into contact with the seed crystal, which is then pulled upwards so that it grows downwards, with the diameter being adjusted to approximately 3 mm to 6 mm. This area of the crystal is called the thin neck (or "dash neck"). The diameter is then further expanded until it reaches the target diameter of the crystal.
[0042] Both the crucible and the single crystal can be rotated. The directions of rotation are generally opposite to each other. This rotation is intended, for example, to obtain a substantially circular cylindrical shape for the single crystal.
[0043] Once the target diameter is reached, a substantially cylindrical crystal with that target diameter is grown. Naturally, minor diameter variations occur during this process, which are due to the diameter control measures. These variations are typically less than 0.5 mm.
[0044] After the cylindrical part of the crystal has been drawn, the diameter is usually reduced - as described, for example, in EP 3 411 515 A1 - and then separated from the melt, whereby the resulting end cone is not used for the further manufacturing process of semiconductor devices, i.e., it is discarded.
[0045] After considerable experimental effort, the inventors discovered that it is possible to draw the monocrystalline rod from the melt even without an explicit end cone, without causing tempering or dislocations. Dislocation- and slip-free is defined as the presence and visibility of the drawing edges of the resulting crystal along its entire length. A crystal forms drawing edges (habit lines) around its circumference because the material grows faster in certain crystal directions than in others. Single crystals of silicon, which are in a <100> -direction, for example, forms four drawing edges at intervals of 90° each.
[0046] The shape of the end piece required after drawing the cylindrical section of the crystal is defined by the process flow at the end of the cylindrical section and is described below, one advantage of which is 202400024 / Kl
[0047] 9. The length of the end piece is significantly shorter than the length of a conventionally drawn end cone. The length of the end piece is defined as the distance between the end of the substantially cylindrical section of the rod and the end of the end piece. This is illustrated in Figures 2, 3, and 4.
[0048] The resulting advantages include a reduction in the loss of semiconductor material and, at the same time, a drastic reduction in plant time.
[0049] According to the invention, in the process a substantially cylindrical section of the single crystal with a target diameter is drawn at a medium drawing speed and crystal rotation using a barrel heater with medium electrical power and preferably using a CUSP magnetic field.
[0050] To complete the production of the substantially cylindrical section of the single crystal, the drawing speed is reduced to less than 15%, preferably 10%, particularly preferably 5%, of the previous average drawing speed, and the power of the barrel heater is increased by at least 15% to 40% compared to the previous average power.
[0051] Preferably, a crucible reversal rotation is used simultaneously, wherein the magnitude of the rotational speed averaged over time is less than 1 rpm, and wherein the direction of rotation is continuously changed, and the amplitude of the rotational speed before and after the change of direction of rotation is not less than 0.5 rpm and not more than 3.0 rpm.
[0052] According to the invention, the single crystal is separated from the melt after a minimum of 15 minutes and a maximum of 60 minutes.
[0053] Preferably, the crucible is moved downwards during separation in order to separate the single crystal from the melt.
[0054] Preferably, a ring heater is used when drawing the substantially cylindrical section of the crystal, the electrical power of which is preferably increased by at least 0.5 kW up to a maximum of 5 kW at the end of the drawing process. 202400024 / Kl
[0055] 10
[0056] The inventive method yields a silicon crystal rod comprising a substantially cylindrical section with a nominal diameter and an adjoining end piece with a length L, wherein the length L of the end piece is less than 10% of the nominal diameter. A length L of less than 6% of the nominal diameter is particularly preferred.
[0057] The end piece preferably contains a convex surface with a point in the middle.
[0058] Preferably the nominal diameter is larger than 280 mm and smaller than 320 mm.
[0059] Preferably, the resulting crystal rod is free of dislocations.
[0060] Preferably, the end piece contains dopant variations (also called striations) whose shape runs parallel to the surface near the surface.
[0061] If a rod produced using the inventive method is cut lengthwise (i.e. axially), a so-called board can be obtained with a width corresponding to the nominal diameter and a length corresponding to the rod.
[0062] Measurements can be taken on this board that characterize both the crystal and the drawing process used to produce the crystal.
[0063] The dopant added to the melt in the CZ process incorporates irregularly into the crystal. Preferably, the dopant contains boron or phosphorus. This leads to a locally inhomogeneous resistivity distribution of the silicon, which is called "striation".
[0064] Although great efforts are made to avoid striations in order to prevent negative effects during the component manufacturing process, striations are still usually measurable once the drawn rod has been doped using the CZ process.
[0065] Since the dopant is incorporated from the melt into the crystal along the melt / crystal interface, the original shape of the interface between the crystal and the substrate can be determined by analyzing the measured resistance distribution.
[0066] 11
[0067] Melt formation can be determined by observing growth lines. Two examples of literature addressing this measurement and evaluation method are listed below:
[0068] Investigation of defects and striations in as-grown Si crystals by SEM using Schottky diodes, Appl. Phys. Lett. 27, 313 (1975); https: / / doi.Org / 10.1063 / 1.88482, AJR de Kock, SD Ferris, LC Kimerling, and HJ Leamy and
[0069] Lüdge, A., Riemann, H.: Doping in homogeneities in silicon crystals detected by the lateral photovoltage scanning (LPS) method. Inst. Phys. Conf. Ser. 160, 145-148 (1997).
[0070] The latter source (Lüdge et al.) describes the method of “lateral photovoltaic scanning” (LPS), which is also suitable for reconstructing the interface between crystal and melt, i.e. the growth lines, when the resistance set by doping is high, i.e. the dopant concentration is low.
[0071] When the method of "lateral photovoltaic scanning" (LPS) is applied to the board described above, contours of the growth lines can be determined, which accurately reflect the deflection of the interface between the melt and the crystal.
[0072] The inventors have discovered that the end piece of the crystal produced according to the inventive method contains dopant variations whose shape runs parallel to the surface near the surface. This contrasts with a crystal piece produced according to a prior art method, in which the striations intersect the said surface at an angle. Near the convex surface of the end piece, these striations run parallel to this surface, the term "near" preferably being understood as a distance of 5 mm.
[0073] When striations are examined at a greater distance from the convex surface, it is noticeable that the shape of the striations changes from convex to concave in the radial center of the crystal. A greater distance is defined here as 202400024 / Kl
[0074] 12 approximately 5 cm - 8 cm. The shape of the striation can be plotted in a coordinate system. The magnitude of the deflection can be determined by subtracting the minimum value of the curve from the maximum value of the curve and taking the absolute value of the result. Preferably, the measured deflection is less than 10 mm.
Claims
202400024 / Kl 13 Patent claims 1. Czochralski method for growing a single crystal from silicon, comprising the following steps: (a) the melting of polycrystalline silicon in a crucible to a melt; (b) bringing a seed crystal into contact with the surface of the melt; (c) drawing the seed crystal into a thin neck with a diameter; (d) widening the diameter of the thin neck to a target diameter; (e) the drawing of a substantially cylindrical section of the single crystal with said target diameter at a nominal drawing speed and crystal rotation using a drum heater with an electrical power (f) the cessation of the drawing of the substantially cylindrical section; (g) the separation of the single crystal from the melt, characterized in that, upon completion of the drawing of the substantially cylindrical section, the drawing speed is reduced to less than 15% of the nominal drawing speed, the power of the drum heater is increased by at least 15% to 40%, and after a minimum of 15 min to a maximum of 60 min after the commencement of completion of the drawing of the substantially cylindrical section, the single crystal is separated from the The melt is separated.
2. Method according to claim 1, characterized in that the drawing of the substantially cylindrical subsection is carried out using a CUSP magnetic field.
3. Method according to claim 1, characterized in that the crucible is moved downwards to separate the single crystal from the melt.
4. Method according to claim 1, characterized in that the drawing speed is reduced to less than 10%, preferably less than 5%, of the nominal drawing speed. 202400024 / Kl 14 5. Method according to claim 1, characterized in that a ring heater is used to heat the crystal when drawing the substantially cylindrical section of the crystal.
6. Method according to claim 1, characterized in that a ring heater is used when drawing the substantially cylindrical section of the crystal, the electrical power of which is increased by at least 0.5 kW up to a maximum of 5 kW when the drawing of the substantially cylindrical section is completed.
7. Silicon crystal rod comprising a substantially cylindrical section with a nominal diameter and an adjoining end piece with a length L, characterized in that the length L of the end piece is less than 20% of the nominal diameter and the end piece has a substantially convex surface and a tip in the middle of the convex surface and the end piece contains dopant variations whose shape is parallel to the surface near the surface.
8. Crystal rod according to claim 7, characterized in that the length L of the end piece is less than 10% (preferably 6%) of the nominal diameter.
9. Crystal rod according to claim 7, characterized in that the nominal diameter is greater than 280 mm and less than 320 mm.
10. Crystal rod according to claim 7, characterized in that the crystal rod is dislocation-free.
11. Crystal rod according to claim 7, characterized in that the deflection is less than 10 mm.