Method and apparatus for producing defect-free single-crystal silicon crystals

Abstract

The present invention relates to a method and apparatus for manufacturing defect-free single-crystal silicon crystals. A crystal puller apparatus includes a pulling assembly for pulling a crystal from a silicon melt at a pulling speed; a crucible for containing the silicon melt; a heat shield above the surface of the silicon melt; an elevator for changing the gap between the heat shield and the surface of the silicon melt; and one or more computing devices for determining an adjustment to the gap using a Pv-Pi tolerance at a given length of the crystal in response to a change in the pulling speed. A computer-implemented method performed by the computing device includes determining a pulling speed command signal to control the diameter of the crystal; determining an elevator command signal to control the gap between the heat shield and the surface of the silicon melt from which the crystal is grown; and determining an adjustment to the gap using a Pv-Pi tolerance in response to different pulling speeds.

Description

Background Art

[0001] The Czochralski (CZ) process was developed to create single crystal or single crystalline material. One of the most important applications of the CZ process in the modern computing world is the growth of single silicon crystals, which are sliced ​​into silicon wafers for the manufacture of semiconductor circuits. Briefly described, the CZ process involves melting a charge of polycrystalline silicon in a quartz crucible and rotationally pulling a seed crystal from the surface of the silicon melt. As the seed crystal is pulled from the silicon melt, the single crystal of silicon extends from the seed crystal and forms a columnar crystal. Modern CZ processes are capable of producing silicon crystals with a diameter of 450 mm or more.

[0002] Figure 1 The typical silicon crystal 100 shown in has a shoulder 102 formed at the beginning of the process and connected to the seed crystal, a body 104 that can extend to 2 meters or more in length, and a tail 106 formed at the end of the process. For the purpose of commercial applications and product yield, it is important to grow crystals with uniform diameters in the body. It is well known to those skilled in the art that increasing the pulling speed of the seed crystal tends to reduce the diameter of the crystal, and thus reducing the pulling speed tends to increase the diameter. However, accurate control of the diameter for silicon crystals often requires complex feedback control systems that also involve factors other than the pulling speed. Summary of the invention

[0003] In one embodiment, a crystal puller apparatus includes: a pulling assembly for pulling a crystal from a silicon melt at a pulling speed; a crucible containing the silicon melt; a heat shield above the surface of the silicon melt; an elevator for changing the gap between the heat shield and the surface of the silicon melt; and one or more computing devices for determining an adjustment to the gap using a Pv-Pi margin at a given length of the crystal in response to a change in the pulling speed.

[0004] In one embodiment, the Pv-Pi tolerance includes a first boundary for the Pv region and a second boundary for the Pi region.

[0005] In one embodiment, the Pv-Pi tolerance also includes a center tolerance halfway between the first boundary and the second boundary.

[0006] In one embodiment, the one or more computing devices use the center tolerance to determine the adjustment to the gap.

[0007] In one embodiment, the pull speed corresponds to a crystal growth rate (v) of a Voronkov ratio (v / G), and the gap constitutes a temperature gradient value (G) of the Voronkov ratio (v / G).

[0008] In one embodiment, one or more of the computing devices maintain the Voronkov ratio at a desired value by adjusting the gap in response to any changes in pull speed.

[0009] In one embodiment, a crystal puller apparatus includes: a pulling assembly for pulling a crystal having a desired diameter from a silicon melt according to a pulling speed profile providing a pulling speed value; a crucible for accommodating the silicon melt; a heat shield above the surface of the silicon melt; an elevator below the crucible for vertically moving the crucible to control the gap between the heat shield and the surface of the silicon melt according to a gap profile providing a gap value; a measuring device for measuring an actual diameter of the crystal ingot; and one or more computing devices for changing the pulling speed value to provide different pulling speeds so as to maintain the actual diameter of the crystal ingot at the desired diameter, and for further adjusting the gap value using a Pv-Pi tolerance in response to different pulling speeds to provide gap adjustment.

[0010] In one embodiment, the Pv-Pi tolerance includes a first boundary for the Pv region and a second boundary for the Pi region.

[0011] In one embodiment, the Pv-Pi tolerance also includes a center tolerance halfway between the first boundary and the second boundary.

[0012] In one embodiment, one or more computing devices use a center tolerance to determine the gap adjustment.

[0013] In one embodiment, each of the pull speed values ​​of the pull speed profile corresponds to a crystal growth rate (v) of a Voronkov ratio (v / G), and each of the gap values ​​of the gap profile constitutes a temperature gradient value (G) of the Voronkov ratio (v / G).

[0014] In one embodiment, one or more of the computing devices maintain the Voronkov ratio at a desired value by adjusting the gap value in response to any changes in the pull speed value.

[0015] In one embodiment, one or more computing devices perform active temperature gradient control using a center tolerance, a pull speed profile, and a gap profile.

[0016] In one embodiment, active temperature gradient control is based on different pull speeds and uses a center tolerance, a pull speed profile, and a gap profile to determine the gap adjustment.

[0017] In one embodiment, the elevator moves the crucible vertically.

[0018] In one embodiment, a computer-implemented method of growing a crystal includes determining, by a computing device, a pull speed command signal to control a diameter of the crystal; determining, by the computing device, an elevator command signal to control a gap between a heat shield and a surface of a silicon melt from which the crystal is grown; and determining, by the computing device, an adjustment to the gap using a Pv-Pi tolerance in response to different pull speeds.

[0019] In one embodiment, the target gap is based on a predetermined function of pull rate relative to gap at different crystal lengths, wherein the predetermined function is within a Pv-Pi tolerance that defines a range of acceptable v / G ratios for growing substantially defect-free crystals, where v is the pull rate of the crystal and G is the temperature gradient at the solid-liquid interface of the crystal.

[0020] In one embodiment, the elevator command signal is based on a comparison of the gap measurement to a target gap.

[0021] In one embodiment, the pull speed command signal is based on a comparison of a measured diameter of the crystal to a diameter profile value that provides a pull speed correction value, and the pull speed command signal is based on a comparison of the pull speed correction value to the pull speed profile value.

[0022] In one embodiment, the diameter profile value is determined based on a function of crystal diameter relative to crystal length, and the pull speed profile is determined based on a function of pull speed relative to crystal length.

[0023] In one embodiment, the pull speed command signal is used to calculate the target gap.

[0024] This Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter.

[0025] In this specification, reference to "an example" or "an embodiment" or "an embodiment" means that a particular feature, structure, or characteristic is included in at least one example or embodiment. Thus, the phrases "in one example" or "in an embodiment" or "in an embodiment" do not necessarily all refer to the same example or embodiment. Furthermore, in one or more examples or embodiments, the particular features, structures, or characteristics may be combined in any suitable manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The foregoing aspects and many of the attendant advantages of the present invention will become more readily appreciated as they become better understood by reference to the following detailed description when considered in conjunction with the accompanying drawings, wherein:

[0027] Figure 1 It is a representative shape of crystals;

[0028] Figure 2 is a graphical representation of defective and defect-free regions of a crystal as a function of the Voronkov ratio (v / G);

[0029] Figure 3 is a schematic illustration of a crystal puller apparatus according to an embodiment of the present disclosure;

[0030] Figure 4 is a schematic diagram of a computing device according to an embodiment of the present disclosure;

[0031] Figure 5 is a schematic diagram of a computing device according to an embodiment of the present disclosure;

[0032] Figure 6 is a process control loop diagram according to an embodiment of the present disclosure;

[0033] Figure 7 is a graph illustrating various profiles as a function of crystal length according to an embodiment of the present disclosure;

[0034] Figure 8 is a diagram of an active temperature gradient control loop according to an embodiment of the present disclosure;

[0035] Fig. 9 is a graph illustrating a center tolerance profile as a function of pull speed and gap, the upper and lower ranges as a function of pull speed and gap defining an acceptable range of Pv-Pi tolerance;

[0036] Fig.10 is a flow chart of a computer-implemented method for determining a target gap between a thermal shield and a solid-liquid interface;

[0037] Fig.11A is a representative graph showing fluctuations in actual pulling speed compared to the speed command signal;

[0038] Fig. 11B is a representative graph showing the relationship of v / G as a function of distance from the center of the crystal;

[0039] Fig. 11C It shows that when the gap changes Fig. 11B A representative graph of the changes in the relationship between

[0040] Fig.11D is a representative graph showing the relationship of v / G as a function of distance from the center of the crystal;

[0041] Fig.11E It shows that when the gap changes Fig.11D a representative graph of changes in the relationship between

[0042] Fig.12 Comparison of yield loss between simulation results of the embodiment of the present invention and a comparative example is shown. DETAILED DESCRIPTION

[0043] Conventionally, the CZ process aims to maintain the diameter of the silicon crystal by adjusting the pull rate and the temperature of the silicon melt in the crucible. However, it is well known that variations in the pull rate performed to control the crystal diameter have a negative impact on the defect distribution within the crystal. It is also well known that during the growth of heavily doped crystals, pull rate variations have a negative impact on morphological stability. Voronkov pointed out that the grow-in defect distribution is a solid-liquid interface 350 ( Figure 3 ) is a function of the ratio (v / G) between the growth rate (or crystal pulling rate) v (mm / min) and the axial temperature gradient G (degrees Celsius / mm) (called the Voronkov ratio). See VV Voronkov, J. Crystal Growth, 1982, 59, 625.

[0044] Figure 2 is a vertical cross-section of a single-crystal silicon crystal (hereinafter referred to as "crystal," "single crystal," "silicon crystal," "ingot," "ingot," or "silicon ingot") 200, illustrating types of growth defects and their distribution. Figure 2 The effect on crystal growth when gradually reducing the pull speed v and thus reducing the Voronkov ratio is shown.

[0045] When the Voronkov ratio (v / G) is large, a collection of vacancies (or vacancy defects) called COPs (Crystal Originated Particles) is formed in the crystal. COPs are crystallographic defects, which are lattice sites that silicon atoms would occupy in a perfect crystal but are missing. On the other hand, when the Voronkov ratio is small, dislocation clusters called L / DLs (large dislocation loops) are formed in the crystal. Dislocation clusters are aggregations of interstitial silicon that are excessively bound between the crystal lattices. L / DLs also represent crystallographic defects caused by interstitial defects of silicon atoms that occupy sites where atoms are not normally present in the crystal structure. Oxidation-induced stacking faults (OSFs) are harmful recurrent defects that occur along silicon crystals obtained by the CZ method. COPs, L / DLs, and OSFs can degrade the performance of integrated semiconductor devices manufactured from wafers cut from such defective silicon crystals.

[0046] On the other hand, crystal 200 includes an oxygen precipitation promoting region or vacancy point defect region (Pv region) and an oxygen precipitation inhibiting region or interstitial point defect region (Pi region) between the OSF region and the L / DL region. COP and L / DL are not present in the Pv region and the Pi region. The Pv region is considered to be a defect-free region in which vacancy type point defects are dominant, and the Pi region is considered to be a defect-free region in which interstitial silicon type point defects are dominant. Point defects are considered to be non-defects due to their size and because they do not negatively affect the performance of devices manufactured from wafers cut from the silicon crystal. Therefore, the key to manufacturing defect-free silicon crystals and wafers manufactured therefrom is to maintain the Voronkov ratio (v / G) within the Pv and Pi regions when growing the silicon crystal, and the Pv and Pi regions are referred to as Pv-Pi tolerances and are described in detail in the accompanying drawings. Figure 2 is delineated by boundaries A and B.

[0047] According to the present disclosure, a computer-implemented method and system provides a process for growing a bulk (104, Figure 1 ) with a crystal having a substantially constant diameter and thereby obtaining a defect-free wafer.

[0048] According to an embodiment of the present disclosure, a heat shield above the silicon melt is used to control the temperature gradient at the solid-liquid interface. Specifically, the distance between the bottom of the heat shield and the silicon melt (i.e., the gap or gap distance) can be changed to control the amount of heat flow that affects the axial temperature gradient of the crystal at the solid-liquid interface. Therefore, throughout the present disclosure, controlling the gap constitutes controlling the temperature gradient of the crystal at the solid-liquid interface.

[0049] I. Pulling machine equipment

[0050] Figure 3 A representative example of a crystal CZ puller apparatus 300 (or puller) that can be used to implement the systems and methods of the present disclosure is shown. The illustrated puller 300 is a simplified schematic diagram that omits illustration of some components of a conventional puller because their inclusion would only obscure some aspects of the present disclosure and their omission does not hinder understanding. Generally speaking, a puller is used to grow silicon crystals by the Czochralski method.

[0051] In one embodiment, the puller 300 includes an exterior enclosure 302 that can be used to perform crystal growth in an inert atmosphere such as argon. In one embodiment, the puller 300 includes a quartz crucible 304 to hold a silicon melt 306. In one embodiment, the quartz crucible 304 is a bowl-shaped container having upright tubular walls and a closed bottom and an open top.

[0052] In one embodiment, the quartz crucible 304 is received within a carbonaceous crucible 308. Like the quartz crucible 304, the carbonaceous crucible has a tubular wall, a closed bottom, and an open top. The inner diameter of the carbonaceous crucible 308 is equal to or slightly larger than the outer diameter of the quartz crucible 304, and the closed bottom of the quartz crucible 304 is seated on top of the closed bottom of the carbonaceous crucible 308.

[0053] In one embodiment, the carbonaceous crucible 308 is placed on the end of a drive shaft 310. A lifter 312 drives the drive shaft 310 to rotate and move the carbonaceous crucible 308 and thereby move the quartz crucible 304 vertically. The lifter 312 can include an electric servo motor having a shaft connected to a gear reducer, which in turn is connected to the drive shaft 310. The electric motor can have a variable frequency drive to control the rate at which the carbonaceous crucible 308 is lifted and for controlling the direction to either lift or lower the carbonaceous crucible 308. In other embodiments, the lifter 312 can be configured differently to provide an alternative mechanism for lifting the quartz crucible.

[0054] The puller 300 includes a carbon heater 314 that surrounds the carbonaceous crucible 308 and is used to provide heat to the quartz crucible 304 to control the temperature of the silicon melt 306 in the quartz crucible 304 .

[0055] In one embodiment, a tubular radiant heat shield 316 is located above the silicon melt 306. In one embodiment, the heat shield 316 reduces the amount of heat that reaches the crystal 318. In one embodiment, the heat shield 316 can be made of metal, graphite, ceramic, or other materials, and the thickness can vary depending on the material.

[0056] In one embodiment, the heat shield 316 has a flat annular section 320 that is located directly above and approximately parallel to the silicon melt 306. The annular flat section 320 has an orifice 348 in the center to allow the crystal 318 to be pulled through the orifice 348. The outer diameter of the flat annular section 320 connects to a constant diameter tubular section 322. The outer diameter of the tubular section 322 is less than the inner diameter of the quartz crucible 304, so that all of the flat annular section 320 and a portion of the tubular section 322 are within and below the upper edge of the top opening of the quartz crucible 304. The tubular section 322 terminates at the top with a connection to a second annular section 324, where the tubular section 322 connects to the inner diameter of the orifice in the second annular section 324. The outer diameter of the second annular section 324 connects to a tapered section 326 that is ultimately secured to the interior of the enclosure 302 of the puller 300. In other embodiments, the heat shield can be configured differently. In an embodiment, the heat shield 316 can be configured with a lifter to raise and lower the heat shield 316. The lifter for the heat shield can be an alternative or supplement to the lifter 312 for the quartz crucible.

[0057] In one embodiment, the vertical distance (i.e., gap 340) between the bottom side 342 of the first annular section 320 of the heat shield 316 and the melt surface 328 can be adjusted by the lifter 312 through the vertical movement of the drive shaft 310 and the carbonaceous crucible 308. The bottom side 342 of the heat shield 316 is the side facing the melt surface 328. This adjustment of the gap 340 changes the temperature gradient G of the crystal 318 at the solid-liquid interface 350 and is used to control the Voronkov ratio within the Pv-Pi tolerance according to the present disclosure.

[0058] The puller 300 includes a pull assembly 334 located approximately on top of the enclosure 302. The pull assembly 334 includes a servo motor 346 and a pulley attached to a cable 330, which extends vertically downward and is aligned approximately concentrically with the center of the orifice 348 of the first annular section 320 of the thermal shield 316. The lower end of the cable 330 is attached to the seed crystal 332. The pull assembly 334 pulls the seed crystal 332 vertically upward from the silicon melt 306 by means of the cable 330 via the servo motor 346. Figure 3 Crystal 318 is shown growing from the end of seed crystal 332 as servo motor 346 pulls cable 330 .

[0059] The puller 300 includes an optical device 336 (e.g., a CCD camera) that is used to measure the diameter of the crystal 318 through a window 338 of the outer enclosure 302. The optical device 336 can also be used to measure the gap 340 between the bottom side 342 of the heat shield 316 and the melt surface 328 of the silicon melt 306.

[0060] Figure 3 A computing system 344 configured for process control of the CZ puller 300 is shown. The computing system 344 is electrically connected to send or receive control signals from at least the elevator 312, the pulling assembly 334, and the heater 314. The computing system 344 also receives measurement signals from at least the optical device 336 to measure the crystal 318 diameter and the gap 340. Other measurements may include one or more temperatures from one or more locations on the CZ puller 300. The process control loop is further described below.

[0061] II. Computing System

[0062] Figure 4 4 is a schematic illustration of a computing system 344. In one embodiment, the computing system 344 can include a central computing server or a personal computer or multiple computers 410 connected via a network (such as the Internet or an intranet or any local area network) that executes the software components illustrated in box 416. The computing device 410 includes one or more processors, non-transitory computer-readable media, and a network communication interface, wherein the network communication interface allows communication with each other and with the central computing server in a distributed computing environment. The computing device 410 can also receive and transfer data through a transient computer-readable medium (such as a USB drive).

[0063] In one embodiment, computing system 344 includes an engine configured to perform a computer-implemented method or routine. "Engine" refers to logic implemented in hardware or software instructions, which can be written in programming languages ​​such as C, C++, COBOL, JAVA™, PHP, Perl, HTML, CSS, JavaScript, VBScript, ASPX, Microsoft .NET™, Go, etc. The engine can be compiled into an executable program or written in an interpretive programming language. The software engine can be called from other engines or from itself. Generally, the engine described herein refers to a logic module that can be merged with other engines or can be divided into sub-engines. The engine can be stored in any type of computer-readable medium or computer storage device, and can be stored on and executed by one or more general-purpose computers, so a special-purpose computer configured to provide the function of such an engine is created.

[0064] In one embodiment, the computing system 344 is configured to execute at least a plurality of process control loops. To this end, the computing system 344 may include: (a) a diameter control engine 406 configured to control the pull speed of the crystal 318 based on the measured diameter, the diameter profile, and the speed profile so as to produce a desired crystal diameter; (b) a gap control engine 402 configured to vertically control the elevator 312 based on the measured gap and the target gap value; and (c) an active temperature gradient control engine (AGC) 404 that calculates a target gap value based on the pull speed command signal, the pull speed profile, the gap profile, and the tolerance profile in response to any change in the crystal pull speed (v) to achieve a desired temperature gradient value (G) such that the v / G ratio is maintained within the Pv-Pi tolerance. Note that the basic functionality of the active temperature gradient control described in other embodiments is the same as the basic functionality of the active temperature gradient control engine (AGC) 404 in this embodiment.

[0065] "Data store" refers to any suitable device that is configured to store data for access by a computing device or engine. An example of a data store is a highly reliable, high-speed relational database management system (DBMS) that is executed on one or more computing devices and can be accessed over a high-speed network. Another example of a data store is a key-value store. However, any other suitable storage technology and / or device that can quickly and reliably provide stored data in response to queries can be used, and the computing device can be locally accessible rather than accessed over a network, or can be provided as a cloud-based service. The data store may also include data stored in an organized manner on a computer-readable storage medium, such as a hard drive, flash memory, RAM, ROM, or any other type of computer-readable storage medium. Those of ordinary skill in the art will recognize that the separate data stores described herein may be combined into a single data store, and / or the single data store described herein may be divided into multiple data stores without departing from the scope of the present disclosure.

[0066] In one embodiment, data store 408 is used to store configuration files containing values ​​to be used for at least gap control engine 402, diameter control engine 406, and AGC 404. In one embodiment, configuration files can refer to functions, tables, or relationships of one variable to one or more variables, etc. For example, functions of pull speed to crystal length, diameter to crystal length, gap to crystal length, and pull speed to gap can be predetermined by experimental and / or theoretical methods and stored in data store 408.

[0067] Figure 5 is a block diagram illustrating aspects of an exemplary computing device 410 .

[0068] In its most basic configuration, computing device 410 includes at least one processor 502 and system memory 504 connected by a communication bus 506. Depending on the exact configuration and type of computing device, system memory 504 can be volatile or non-volatile memory, such as read-only memory ("ROM"), dynamic random access memory ("DRAM"), EEPROM, flash memory, or similar memory technology. Those of ordinary skill in the art and others will recognize that system memory 504 typically stores data and / or program modules that are immediately accessible to and / or currently being operated on by processor 502. Processor 502 serves as the computational center of computing device 410 by supporting the execution of instructions.

[0069] like Figure 5 As further illustrated in , computing device 410 may include a network interface 510 that includes one or more components for communicating with other devices over a network. Embodiments of the present disclosure may access basic services that utilize network interface 510 to perform communications using common network protocols. Network interface 510 may also include a wireless network interface that is configured to communicate via one or more wireless communication protocols such as WiFi, 2G, 3G, LTE, WiMAX, Bluetooth, Bluetooth low energy, etc. As will be appreciated by one of ordinary skill in the art, Figure 5 The network interface 510 illustrated in FIG. 4 may represent one or more wireless interfaces or physical communication interfaces described and illustrated above with respect to specific components of the computing device 410 .

[0070] exist Figure 5 In the exemplary embodiment depicted in , computing device 410 also includes storage medium 508. However, the service may be accessed using a computing device that does not include a device for persisting data to a local storage medium. Figure 5 The storage medium 508 depicted in the figure may be optional. In any case, the storage medium 508 may be volatile or non-volatile, removable or non-removable, implemented using any technology capable of storing information (such as, but not limited to, hard disk drives, solid-state drives, CD ROMs, DVDs or other disk storage, cassettes, magnetic tapes, disk storage, etc.).

[0071] As used herein, the term "computer-readable media" includes volatile and nonvolatile, and removable and non-removable media implemented in any method or technology capable of storing information such as computer-readable instructions, data structures, program modules or other data. In this regard, Figure 5 The system memory 504 and storage media 508 depicted in FIG. 5 are merely examples of computer-readable media.

[0072] Suitable implementations of computing device 410, including processor 502, system memory 504, communication bus 506, storage media 508, and network interface 510, are known and commercially available. For simplicity of illustration and because they are not important to understanding the claimed subject matter, Figure 5 Some typical components of many computing devices are not shown. In this regard, computing device 410 may include input devices (such as a keyboard as shown), and may also include a keypad, a mouse, a microphone, a touch input device, a touch screen, a tablet computer, and the like. Such input devices may be coupled to the computing device via a wired or wireless connection, including RF, infrared, serial, parallel, Bluetooth, Bluetooth low energy, USB, or other suitable connection protocols using wireless or physical connections. Similarly, computing device 410 may also include output devices (such as a display as shown), and may also include speakers, printers, and the like.

[0073] III. Crystal diameter, crystal length and gap distance measurements

[0074] In one embodiment, Figure 4 As shown in , diameter control engine 406 uses a measurement of the diameter of crystal 318 at about solid-liquid interface 350 , and gap control engine 402 uses a measurement of gap 340 between the top of melt surface 328 and bottom side 342 of heat shield 316 .

[0075] In one embodiment, the optical device 336 can be used to measure both the crystal diameter and the gap distance. In one embodiment, the crystal diameter is measured according to the technology known from U.S. Patent No. 8,414,701. In one embodiment, the gap distance is measured according to the technology known from U.S. Patent No. 9,567,692. None of these measurement techniques need to be explained in detail in order to understand the systems and methods of the present disclosure, as reference can be made to the aforementioned patents.

[0076] Embodiments of process control in the present disclosure may use a measurement of the current crystal 318 length to determine values ​​from the diameter profile, pull speed profile, gap profile, and tolerance profile. In one embodiment, the crystal length is the length from the shoulder 102 ( Figure 1) to the solid-liquid interface 350. In one embodiment, for example, the length of crystal 318 is determined by an encoder that counts the rotations of the motor of servo motor 346. The encoder can then convert the number of rotations into a corresponding length of crystal 318. In one embodiment, the length of crystal 318 is determined based on the weight and density of crystal 318 and the geometry or shape. In one embodiment, the length of crystal 318 can be directly measured by an optical device (such as, optical device 336).

[0077] IV. Process Control Model

[0078] Figure 6 An embodiment of a process control loop for a CZ puller 300 is shown. The process control loop can use measured variables, input set point variables from a configuration file, and then manipulate the variables to give the desired process output. Process disturbances (such as inaccuracies in the actual speed of the seed crystal 332) can affect the process output. In one embodiment, for example, the present disclosure addresses process disturbances by controlling the v / G ratio within the Pv-Pi tolerance to minimize defects in the crystal due to variations in the actual speed.

[0079] exist Figure 6 , the measured variables from the optical device 336 include the crystal diameter measurement (Dm) and the gap measurement (Gm). It should be understood that the crystal length is also measured or calculated by indirect means. The command variables include a pull speed command signal (Scom) sent to the pull assembly 334, which affects the diameter of the crystal 318; and a lift command signal (Lcom) sent to the lifter 312, which changes the gap 340 between the melt surface 328 and the heat shield 316. The pull speed command signal (Scom) is the output from the diameter control engine 406, and the lift command signal (Lcom) is the result of the AGC 404 and the gap control engine 402. In this article, the command signal represents the target output desired by the controlled device. The command signal can represent a numerical value, such as mm / min. In a physical sense, the command signal can be, for example, an electrical signal of a voltage change or a pneumatic signal.

[0080] In one embodiment, the gap control engine 402 is controlled in tandem with the AGC 404, while the AGC 404 is controlled in tandem with the diameter control engine 406. In other words, the output from the diameter control engine 406 is used in the AGC 404, and the output from the AGC is used in the gap control engine 402. The AGC 404 and the gap control engine 402 adjust the gap 340. Herein, the command signals (Scom and Lcom) represent the desired outputs of the controlled devices including the pull assembly 334 and the elevator 312, respectively.

[0081] The process control loop (i.e., engine) also uses predetermined input values ​​from the diameter profile 602, the pull speed profile 604, the gap profile 606, and the tolerance profile 608. Based on the two measured variables (Dm, Gm), the crystal length, and the two command variables (Scom, Lcom) and the predetermined profiles, the process output is a defect-free crystal of a predetermined diameter. Specifically, the target gap (Gt) is provided to the gap control engine 402 by the AGC 404, and the Voronkov ratio (v / G) is maintained within the Pv-Pi tolerance.

[0082] exist Figure 6 , the relationship between the diameter control engine 406, the gap control engine 402, and the AGC 404 is illustrated. The optical device 336 measures the actual diameter (Dm) of the crystal in real time and feeds this information to the diameter control engine 406. The diameter control engine 406 then calculates the diameter difference (Dd) at block 610 by comparing the measured diameter (Dm) with the diameter value obtained from the diameter profile 602 at the current length of the crystal 318. Based on Dd, the diameter control engine 406 determines the pull speed correction (Sc) at 612. For example, if the diameter difference (Dd) indicates that the actual diameter of the crystal 318 is less than expected, the pull speed correction (Sc) is decreased. Conversely, if the diameter difference (Dd) indicates that the actual diameter of the crystal is greater than expected, the pull speed correction (Sc) is increased.

[0083] At box 612, a conversion is performed to convert Dd in millimeters to Sc in millimeters / minute, for example. At box 614, the diameter control engine 406 then applies Sc to the pull speed input or the value from the pull speed profile (Sp) based on the current crystal diameter to determine the vertical pull speed command signal (Scom) and sends Scom to the pull assembly 334. However, due to Scom, the actual pull speed experienced by the crystal 318 not only affects the diameter, but may also cause disturbances in the Voronkov ratio (v / G) and form defects outside the Pv-Pi tolerance. Therefore, in order to avoid defects, embodiments of the present disclosure may use Scom from the diameter control engine 406 to trigger the AGC 404, which determines the target gap (Gt). More about the operation of the AGC 404 will be discussed in Section 5.

[0084] Then, the gap control engine 402 calculates the lifter command signal (Lcom) and sends Lcom to the lifter 312 to vertically move the crucible 304 to control the gap 340 between the melt surface 328 and the bottom side 342 of the heat shield 316. The gap 340 affects the temperature gradient G at the solid-solution interface 350. Therefore, the gap 340 also affects the defect formation in the crystal. In the present disclosure, a combined control of the pull speed and the gap is used to reduce and / or eliminate defects in the crystal. According to the present disclosure, when the pull speed changes, the AGC 404 calculates the target gap (Gt) to maintain v / G within the Pv-Pi tolerance.

[0085] like Figure 6 As shown in , data storage 408 can include data profiles, including diameter profile 602, pull speed profile 604, gap profile 606, and tolerance profile 608. Profiles can be set as functions, correlations, or relationships, for example, pull speed versus crystal length, diameter versus crystal length, gap versus crystal length, and pull speed versus gap can be predetermined by experimental or theoretical methods. Profiles can also be constructed as multidimensional arrays that have correlations of more than two variables. Profiles can vary based on different sizes of crystals and further due to different pullers.

[0086] In one embodiment, diameter profile 602 provides a set of desired diameter values ​​for a crystal's length (i.e., as a function of the crystal's length). Pull speed profile 604 also provides a set of desired pull speed values ​​for a crystal's length. Gap profile 606 provides a set of desired gap values ​​for a crystal's length. In general, both diameter profile 602 and pull speed profile 604 support Figure 1 In one embodiment, the values ​​may include variations to account for heat transfer, pressure, and other chemical and physical effects of the puller environment, such as variations obtained from simulations and / or test crystal growth.

[0087] The tolerance profile 608 provides a Pv-Pi tolerance at each given length of the crystal. The Pv-Pi tolerance includes (or can be used to calculate) a center tolerance (MC), which is represented as a set of pull speed values ​​and corresponding gap values, which are selected to maintain the Voronkov ratio (v / G) within the Pv-Pi tolerance. In one embodiment, these values ​​are selected by simulation and / or test growth of crystals. In the test growth, each section of the crystal is analyzed, and a determination is performed to identify the pull speed and thermodynamic terms that provide a defect-free crystal zone (i.e., the Pv-Pi tolerance). This data can be compiled and tabulated at various lengths of the crystal. This (tabulated) tolerance profile 608 can be used as a lookup table in the AGC 404 to determine the desired target gap used by the gap control engine 402.

[0088] Figure 7 A graph showing one embodiment of diameter profile values, pull speed profile values, and gap profile values ​​over the length of a silicon crystal according to the present disclosure. In one embodiment, the diameter profile values, pull speed profile values, and gap profile values ​​may be determined by simulation, crystal growth learning experiments, or constitute an average of historical data from multiple experiments on crystals of similar diameter and length. In another embodiment, the averaged historical data provides starting conditions, which then allow for additional fine tuning and selection of more optimized values.

[0089] Figure 6 6 illustrates how the diameter control engine 406 dynamically performs diameter control over the length of the crystal. At adder 610, the diameter control engine 406 compares the value (Dp) from the diameter profile with the diameter measurement (Dm) from the optical device 336 to calculate any diameter difference (Dd). The diameter control engine 406 processes Dd in millimeters, for example, through a converter 612, and determines, for example, a corresponding pull speed correction (Sc) in millimeters per minute required to change the crystal diameter to a pull speed value that satisfies Dd. In one embodiment, the conversion from diameter to speed may be predetermined and compiled in a lookup table. At adder 614, the diameter control engine 406 further compares Sc with the value (Sp) from the pull speed profile at a given length of the crystal to calculate a pull speed command signal (Scom). The diameter control engine 406 then sends the pull speed command signal to the pull assembly 334 (specifically, the servo motor that implements the pull speed command signal (Scom)), which causes the silicon crystal to be maintained at the desired diameter.

[0090] In one embodiment of the diameter control engine 406, instead of using a diameter profile value, the input desired diameter may be set to a single constant value.

[0091] The process control loop uses AGC 404 to prevent defect formation in the crystal 318. AGC 404 calculates a target gap (Gt) in response to a pull speed command signal (ie, Scom) required by the diameter control engine 406. Figure 6 As shown in , the AGC 404 outputs a target gap (Gt) based on a tolerance profile (Mp), a gap profile (Gp), a pull speed profile (Sp), and a pull speed command (Scom).

[0092] The gap control engine 402 uses the target gap (Gt) to determine the elevator command signal Lcom. At the adder 616, the gap control engine 402 compares the target gap (Gt) with the gap measurement (Gm) to generate a gap correction (Gc). The converter 618 converts the gap correction (Gc) in millimeters to a dynamic displacement of the elevator (such as millimeters per minute). This dynamic displacement represents the elevator command signal (Lcom) that vertically moves the elevator 312 to cause a change in the gap 340.

[0093] V. Active Temperature Gradient Control Engine

[0094] Figure 8 One embodiment of an AGC 404 according to the present disclosure is shown. The AGC 404 includes a virtual speed box 802 that uses values ​​from a margin profile (Mp) 608 and a gap profile (Gp) 606 .

[0095] Fig. 9 An embodiment of Mp according to the present disclosure is shown. At a given crystal length, Mp includes a Pv-Pi tolerance, which also includes a center tolerance (MC) selected to be in the middle of the Pv-Pi tolerance. The Pv-Pi tolerance provides a boundary of a vacancy point defect region (Pv region) and another boundary of an interstitial point defect region (Pi region), i.e. Figure 2 The boundary line indicated by AB in Fig. 9 In the example, the Pv-Pi tolerance defines the defect-free area at each of the segments of the crystal. The Mp values ​​can be stored in a three-dimensional array represented by multiple tables (or other data structures, such as graphs, formulas, etc.). These values ​​can be obtained through simulation and / or empirical studies of silicon crystals and their growth parameters. In one embodiment, Fig. 9 As shown in , the center tolerance (MC) is selected to be in the middle of the Pv-Pi tolerance (or in the middle between the Pv boundary and the Pi boundary). This selection allows for equally sufficient margin along either boundary direction of the Pv-Pi tolerance. In another embodiment, a different value that is more biased towards either the Pv region or the Pi region may be used.

[0096] return Figure 8 In one embodiment, the virtual speed block 802 identifies a pull speed value (virtual pull speed (Sv)) corresponding to the current gap value in the gap profile (Gp) 606 from the center tolerance (MC). The virtual speed block 802 performs this task by first retrieving the gap value from the gap profile (Gp) 606 for a given crystal length. The virtual speed block 802 then determines (i.e., looks up) a virtual pull speed (Sv) corresponding to the current gap value of the gap profile (Gp) with reference to the center tolerance (MC) in the tolerance profile 608. Ideally, Sv and the corresponding Gp value provide a Voronkov ratio (v / G) in the defect-free zone or in the Pv-Pi tolerance. This Sv is then used to determine the target pull speed (St) as follows.

[0097] The target speed block 804 calculates a target pulling speed (St) based on the pulling speed command signal (Scom), the virtual pulling speed (Sv), and the pulling speed profile (Sp) at a given crystal length according to the relationship St = Sv * Scom / Sp, where St represents an estimate of the pulling speed resulting from the pulling speed command signal (Scom). Fig. 9 This relationship can also be seen. Here, if Scom = Sp, the ratio Scom / Sp is 1, and the target pull speed St is simply equal to Sv. The ratio (Scom / Sp) is the percentage increase or decrease relative to Sv on the pull speed (Y) axis of the center tolerance (MC) curve.

[0098] The target gap block 806 then determines the target gap (Gt) by referencing the center tolerance in the tolerance configuration file 608 to find the gap value corresponding to the target pull speed (St). The target gap (Gt) is then used by the gap control engine 402 to calculate the gap correction (Gc), as explained above.

[0099] Fig.101 is a flow chart illustrating a process 1000 of the AGC 404 according to another embodiment of the present disclosure. At step 1002, the virtual speed block 802 reads in a tolerance profile (Mp) including a Pv-Pi tolerance. At step 1004, the virtual speed block 802 calculates a center tolerance (MC) based on the Pv-Pi tolerance at a given crystal length. At step 1006, the virtual speed block 802 reads in a gap value from the gap profile (Gp), and at step 1008, the virtual speed block 802 looks up a corresponding virtual pull speed (Sv) from the center tolerance (MC). At step 1010, the target speed block 804 calculates a target pull speed (St), which is an estimate of the pull speed generated by the pull speed command signal (Scom), by taking a ratio of Scom to a pull speed value from the pull speed profile (Sp) as follows: St = Sv * Scom / Sp.

[0100] At step 1012 , the target gap block 806 uses the target pull speed (St) to look up a target gap value (Gt) corresponding to the target pull speed (St) from the center tolerance (MC).

[0101] VI. Simulation Results

[0102] FIG. 11A to FIG. 11E A simulation-assisted inventive example of the AGC 404 according to the present disclosure is illustrated.

[0103] Fig.11A The actual pull speed (measured value) is plotted on the pull speed profile (i.e., system input) at the top of Figure 1. This shows that during crystal growth, the actual pull speed fluctuates from the input profile pull speed. Fig.11A The bottom portion of the graph shows the ratio of the actual pull speed to the profile pull speed to illustrate the extent or sensitivity of pull speed variations.

[0104] Typically, changes in the pull rate required by diameter control affect the Voronkov ratio (v / G) and may cause the crystal to form either vacancy defects or interstitial defects. In order to maintain v / G at a desired level when the speed changes, the temperature gradient (G) also needs to change. The system and method according to the present disclosure changes the gap distance between the bottom side of the heat shield and the top of the silicon melt to achieve a change in the temperature gradient and thereby maintain the ratio (v / G) within the Pv-Pi tolerance (AB) to reduce defects in the crystal.

[0105] Figures 11B to 11E The use of simulations illustrates that adjusting the gap 340 in response to changes in pull speed maintains crystal growth within the defect-free region (AB).

[0106] exist Fig. 11BIn the example above, an increase in the pull rate will increase the Voronkov ratio (v / G) and tend to cause crystal growth (shown as line 1102) into the oxygen induced stacking fault (OSF) region. To compensate for this effect, a corresponding increase in the temperature gradient (G) is required.

[0107] AGC 404 will be used to reduce gap 340, which reduces the amount of heat at the solid-liquid interface and increases G. Fig. 11C As shown in , changing the gap in response to changes in the pull rate results in maintaining the Voronkov ratio (v / G) at a desired level and keeping the crystal growth line 1104 in the defect-free region (AB).

[0108] Similarly, in Fig.11D In the case of , the reduction in pull rate will cause the crystal growth line 1106 to enter the large dislocation (LD / L) region. To compensate for this effect, the AGC 404 will increase the gap, thereby allowing more heat to reach the solid-solution interface and reduce G. Fig.11E As shown in , correction of this gap will again maintain v / G at the desired level to keep the crystal growth line 1108 in the defect-free region (AB).

[0109] VII. Comparison

[0110] Fig.12 A normalized comparison of systems using and not using the active temperature gradient control engine (AGC) 404 of the present invention is illustrated. Here, the lower bar 1202 represents the total wafer loss in the crystal normalized to 1 or 100% from a system not using the AGC 404 of the present invention. Approximately 66% of this loss is attributed to disturbances or fluctuations in the Voronkov ratio (v / G), 1202b. Other factors 1202a (including measurement errors) cause the remaining 34% of the loss. The upper bar 1201 shows the simulated results of wafer loss from the same system using the AGC 404 of the present invention. Bar 1201 shows a significant reduction in wafer loss of approximately 37% overall, all of which is due to the reduction in wafer loss caused by v / G fluctuations. Here, the difference is from 66% to 29%, a net reduction of 37%, and the wafer yield rate has increased by 54% due to the use of the AGC 404 of the present invention.

[0111] While illustrative embodiments have been illustrated and described, it will be appreciated that various changes can be made therein without departing from the spirit and scope of the invention.

Claims

1. A crystal pulling machine device, comprising: A pulling assembly for pulling a crystal from a silicon melt at a pulling speed; A crucible containing the silicon melt; a heat shield above a surface of the silicon melt; a lifter for changing the gap between the heat shield and the surface of the silicon melt; as well as One or more calculation devices that calculate a diameter difference by comparing the measured diameter of the crystal to a desired diameter, change the pull speed based on the diameter difference, and further adjust the gap using a Pv-Pi tolerance in response to the change in the pull speed.

2. The device according to claim 1, wherein: The Pv-Pi tolerance includes a first boundary for the Pv region and a second boundary for the Pi region.

3. The device according to claim 2, wherein: The Pv-Pi tolerance also includes a center tolerance halfway between the first boundary and the second boundary.

4. The device according to claim 3, wherein: The one or more computing devices use the center tolerance to determine an adjustment to the gap.

5. The device according to claim 1, wherein: The pulling speed corresponds to a crystal growth rate (v) of a Voronkov ratio (v / G), and the gap constitutes a temperature gradient value (G) of the Voronkov ratio (v / G).

6. The device according to claim 5, wherein: One or more of the computing devices maintain the Voronkov ratio at a desired value by adjusting the gap in response to any changes in the pull speed.

7. The device according to claim 5, wherein: The adjustment of the gap is determined in such a way that the Voronkov ratio is maintained within the Pv-Pi tolerance.

8. The device according to claim 1, wherein: The elevator moves the crucible vertically.

9. The device according to claim 1, wherein: The Pv-Pi tolerance is provided at various lengths of the crystal.

10. A crystal pulling machine device, comprising: a pulling assembly for pulling a crystal having a desired diameter from a silicon melt according to a pulling speed profile providing a pulling speed value; A crucible containing the silicon melt; a heat shield above a surface of the silicon melt; an elevator below the crucible that vertically moves the crucible to control a gap between the heat shield and a surface of the silicon melt according to a gap profile that provides a gap value; A measuring device for measuring the actual diameter of the crystal; as well as One or more computing devices that calculate a diameter difference by comparing a measured diameter of the crystal to a desired diameter, change one of the pull speed values ​​to provide a different pull speed to maintain the actual diameter of the crystal at the desired diameter, and in response to the different pull speed, further adjust one of the gap values ​​using a Pv-Pi tolerance to provide a gap adjustment.

11. The device according to claim 10, wherein: The Pv-Pi tolerance includes a first boundary for the Pv region and a second boundary for the Pi region.

12. The device according to claim 11, wherein The Pv-Pi tolerance also includes a center tolerance halfway between the first boundary and the second boundary.

13. The device according to claim 12, wherein: The one or more computing devices determine the gap adjustment using the center tolerance.

14. The apparatus according to claim 10, wherein: Each of the pull rate values ​​of the pull rate profile corresponds to a crystal growth rate (v) of a Voronkov ratio (v / G), and each of the gap values ​​of the gap profile constitutes a temperature gradient value (G) of the Voronkov ratio (v / G).

15. The device according to claim 14, wherein: One or more of the computing devices maintain the Voronkov ratio at a desired value by adjusting one of the gap values ​​in response to any change in one of the pull speed values.

16. The apparatus according to claim 12, wherein: The one or more computing devices perform active temperature gradient control using the center tolerance, the pull speed profile, and the gap profile.

17. The device according to claim 16, wherein: The active temperature gradient control determines a gap adjustment based on the different pull speeds and using the center tolerance, the pull speed profile, and the gap profile.

18. A computer-implemented method of growing a crystal, the method comprising: determining a pull speed command signal by a computing device to control the diameter of the crystal, the computing device calculating a diameter difference of the crystal by comparing the measured diameter with a desired diameter, and varying the pull speed based on the diameter difference; determining, by the computing device, a lifter command signal to control a gap between a heat shield and a surface of a silicon melt from which the crystal is grown; as well as The Pv-Pi tolerance is used by the computing device to determine adjustments to the gap in response to different pull speeds.

19. The computer-implemented method of claim 18, wherein: The target gap is based on a predetermined function of pull rate relative to gap at different crystal lengths, wherein the predetermined function is within the Pv-Pi tolerance, which defines a range of acceptable v / G ratios for growing defect-free crystals, where v is the pull rate of the crystal and G is the temperature gradient at the solid-liquid interface of the crystal.

20. The computer-implemented method of claim 19, wherein: The elevator command signal is based on a comparison of the gap measurement to a target gap.

21. The computer-implemented method of claim 18, wherein: Determination of the pull speed command signal includes comparing a measured diameter of the crystal to a diameter profile value to provide a pull speed correction value, and then applying the pull speed correction value to the pull speed profile value.

22. The computer-implemented method of claim 21, wherein: The diameter profile value is determined based on a function of crystal diameter relative to crystal length, and the pull speed profile is determined based on a function of pull speed relative to crystal length.

23. The computer-implemented method of claim 22, wherein: The pulling speed command signal is used to calculate the target gap.

24. The computer-implemented method of claim 23, wherein: The Pv-Pi tolerance includes a first boundary for the Pv region and a second boundary for the Pi region, and a center tolerance halfway between the first boundary and the second boundary.

25. The computer-implemented method of claim 24, wherein: The target gap is determined by: determining a virtual pull speed (Sv) corresponding to a current gap value in a gap profile from the center tolerance, The target pulling speed (St) is calculated based on the pulling speed command signal (Scom), the virtual pulling speed (Sv) and the pulling speed profile (Sp) at a given crystal length according to the relationship St=Sv*Scom / Sp, and the target gap corresponding to the target pulling speed (St) is determined from the center tolerance.

26. The computer-implemented method of claim 25, wherein: The virtual pull speed is the pull speed value from the center tolerance that corresponds to the current gap value from the gap profile at a given crystal length.

27. The computer-implemented method of claim 24, wherein: Determine the target clearance by following these steps: read in the tolerance profile (Mp) including the Pv-Pi tolerance, The center tolerance (MC) is calculated based on the Pv-Pi tolerance at a given crystal length, Read in the gap value from the gap configuration file (Gp), Find the corresponding virtual pulling speed (Sv) from the center tolerance (MC); The target pulling speed (St) is calculated by taking the ratio of the pulling speed command signal (Scom) to the pulling speed value from the pulling speed profile (Sp) as follows: St = Sv*Scom / Sp, and The target gap value (Gt) corresponding to the target pulling speed (St) is found from the center margin (MC) using the target speed (St).

Images