Silicon single crystal drawing method capable of reducing resistivity range of whole silicon single crystal rod

Abstract

The invention relates to a silicon single crystal drawing method capable of reducing the resistivity range of a whole silicon single crystal rod, which belongs to the technical field of wafer processing and comprises the following operation steps of: 1, completely melting a solid silicon material and a doping agent into liquid in a crucible a; and 2, slowly lowering the crucible position in the crucible b to realize that a part of liquid at the bottom of the crucible is slowly solidified into a solid from the bottommost part. And 3, after the crucible in the crucible c is slowly lowered and stopped, seeding, shouldering and equal-diameter operation are carried out. And 4, carrying out an equal-diameter growth stage in the d crucible. And 5, continuously growing the single crystal rod in the e crucible. And 6, finishing silicon single crystal drawing in the crucible f through an ending process. An initial state of'high-concentration melt + low-concentration solid 'is constructed before crystal pulling, and real-time and dynamic dilution compensation is performed on the concentration of a dopant of a main melt, so that the inherent defect of resistivity axial non-uniformity caused by a segregation effect in a traditional Czochralski method is effectively overcome.

Description

Technical Field

[0001] This invention relates to the field of wafer fabrication technology, and specifically to a method for pulling silicon single crystals that reduces the resistivity range of the entire silicon single crystal rod. Background Technology

[0002] The pulling process of single-crystal silicon rods mainly includes the following steps:

[0003] (1) Feeding process:

[0004] Raw material preparation: Carefully selected polycrystalline silicon raw materials are mixed with specific impurities according to N-type or P-type resistance requirements.

[0005] Pour into the crucible: Pour the mixed raw materials into a quartz crucible in preparation for melting and casting.

[0006] (2) Melting process:

[0007] Crystal growth furnace closure: The crystal growth furnace is closed, the vacuum pump starts working, and argon gas is injected to maintain a constant pressure inside the furnace.

[0008] Heating and melting: The graphite heater is ignited, raising the temperature to 1420℃, which melts the polycrystalline silicon raw material.

[0009] (3) Neck-tightening growth:

[0010] Seed crystal immersion: After the melt stabilizes, the seed crystal is immersed into the melt.

[0011] Necking technique: By rapidly lifting the seed crystal, its diameter is reduced to 46mm to reduce dislocations and ensure a flawless crystal surface.

[0012] (4) Shoulder-like growth:

[0013] Adjusting parameters: After the necking process is completed, adjust the temperature and stretching speed simultaneously.

[0014] Diameter expansion: Through shoulder growth technology, the crystal diameter is gradually expanded to the expected size.

[0015] (5) Constant diameter growth:

[0016] Precise control: By precisely controlling the pulling speed and temperature, the diameter of the crystal rod is kept within ±2 mm at the micrometer level.

[0017] Monocrystalline silicon wafer incubation: In this stage, monocrystalline silicon wafers are gradually incubated.

[0018] (6) Tail growth and cooling:

[0019] Tail growth: After the constant diameter section is completed, the diameter of the crystal rod is gradually reduced until a cusp is formed.

[0020] Tail cooling: After tail cooling, the growth cycle of the single crystal silicon rod is completed.

[0021] In conventional silicon single crystal pulling processes, common dopants (boron, phosphorus, arsenic, antimony, gallium, etc.) have segregation coefficients in silicon that are less than 1. This segregation effect results in a wide resistivity distribution across the entire pulled silicon single crystal rod, with a high resistivity at the beginning and a low resistivity at the end. More precisely, with a fixed solidification ratio, the ratio of the highest to the lowest resistivity of the silicon single crystal rod is relatively large. This often leads to lower yields and higher costs because the target product exceeds the required resistivity range. Summary of the Invention

[0022] This invention addresses the shortcomings of existing technologies by providing a method for pulling silicon single crystals that reduces the resistivity range of the entire silicon single crystal rod. This method constructs an initial state of "high-concentration melt + low-concentration solid" before crystal pulling and, during the crystal pulling process, controls the melting rate of the bottom solid to perform real-time and dynamic dilution compensation of the dopant concentration in the main melt. This effectively overcomes the inherent defect of axial non-uniformity of resistivity caused by segregation effect in the traditional Czochralski method.

[0023] The above-mentioned technical problems of the present invention are mainly solved by the following technical solutions:

[0024] A method for pulling silicon single crystals that reduces the resistivity range of the entire silicon single crystal rod includes the following steps:

[0025] Step 1: In crucible a, melt the solid silicon material and dopant together into a liquid.

[0026] Step 2: In crucible b, the crucible is slowly lowered to maintain the heating of the liquid in the upper part of the crucible, reduce the heating of the bottom of the crucible, and increase the heat dissipation from the bottom; thus, a portion of the liquid at the bottom of the crucible slowly solidifies into a solid.

[0027] Step 3: After the crucible in crucible c is slowly lowered and stops, and the liquid to solid ratio in the crucible is stabilized, crystal pulling, shoulder formation, and equal diameter operation are performed.

[0028] Step 4: During the equal-diameter growth stage in crucible d, the heater gradually melts the solid at the bottom of the crucible into a liquid by raising the crucible level.

[0029] Step 5: The single crystal rod continues to grow in crucible e, and the liquid at the top of the crucible grows into a solid single crystal; at the same time, the solid silicon at the bottom of the crucible continues to melt into liquid from top to bottom, becoming a supplement to the liquid silicon material required for the growth of the single crystal rod at the top.

[0030] Step 6: When the single crystal rod grows to the tail end in crucible f, all the solid silicon in the crucible melts into liquid, and the silicon single crystal pulling is completed through the tailing process.

[0031] Preferably, before starting the crystal pulling process, solid silicon is grown at the bottom of the crucible. Since the segregation coefficient of the dopant in silicon is less than 1, the concentration of the dopant in crucible b is increased relative to the concentration in crucible a, and the concentration of the dopant in the solid is decreased.

[0032] Preferably, the dopant concentration before the start of pulling is greater than the dopant concentration in the liquid in the crucible of stage a, and the dopant concentration in the solid is less than the dopant concentration in the liquid in the crucible of stage a. This is because the solid is formed gradually by solidification. Due to the segregation of dopant in silicon, the dopant concentration in the solid is not uniformly distributed, but the dopant concentration at each location is less than the dopant concentration in the liquid in the crucible of stage a.

[0033] Preferably, during the crucible process in steps c, d, and e, the silicon single crystal at the top of the crucible gradually solidifies. Due to segregation, the concentration of dopant in the liquid increases. At the same time, the solid with a lower dopant concentration gradually melts into liquid, which dilutes the dopant concentration in the liquid. This gradual dilution slows down the resistivity change along the length of the single crystal rod. The resistivity of the single crystal rod depends on the dopant concentration, thus reducing the resistivity range of the entire single crystal rod. This reduces the ratio of the highest to the lowest resistivity of the silicon single crystal rod under a fixed solidification ratio.

[0034] Preferably, by controlling the proportion of solid silicon in crucible b to the total amount of silicon, the ratio of the highest resistivity to the lowest resistivity of the silicon single crystal rod can be reduced to varying degrees.

[0035] Preferably, the horizontal axis represents the solidification ratio, which refers to the percentage of the weight of the grown solid silicon single crystal rod to the total feed amount; the vertical axis represents the resistivity of the silicon single crystal rod. When the initial feed amount and doping amount are the same, the resistivity distribution of the silicon single crystal rod is different when different pulling methods are used.

[0036] The solidification ratio is the control knob for the entire process. It represents the percentage of the total feed weight of solid silicon that has solidified at the bottom of the crucible before crystal pulling begins.

[0037] Solidification ratio = 0%: i.e., the traditional Czochralski method. Without any compensation, the resistivity curve shows a steep downward trend from the beginning to the end, with the ρmax / ρ_min ratio at its maximum.

[0038] Solidification ratio > 0%: This method is used. The resistivity curve becomes very flat. The higher the solidification ratio, the more "low-concentration solids" are initially stockpiled, the stronger the subsequent dilution compensation ability, the flatter the resistivity curve, and the smaller the ρmax / ρ_min ratio.

[0039] There is an optimal value: a higher solidification ratio is not always better. An excessively high solidification ratio will result in an excessively high initial main melt concentration (Cl_main), potentially leading to a lower resistivity at the crystal head and increasing the difficulty of process control. Therefore, there exists an optimal solidification ratio range that achieves the best uniformity within the target resistivity range.

[0040] Resistivity: Visually shows the change in resistivity of a single crystal rod along its length (from head to tail).

[0041] Traditional method curve: a sloping line from high to low.

[0042] The curve obtained using this method is a nearly horizontal straight line with minimal fluctuations.

[0043] The present invention can achieve the following effects:

[0044] This invention provides a method for pulling silicon single crystals that reduces the resistivity range of the entire silicon single crystal rod. Compared with the prior art, by constructing an initial state of "high-concentration melt + low-concentration solid" before crystal pulling, and by controlling the melting rate of the bottom solid during crystal pulling, the dopant concentration of the main melt is diluted and compensated in real time and dynamically, thereby effectively overcoming the inherent defect of axial non-uniformity of resistivity caused by segregation effect in the traditional Czochralski method.

[0045] Significantly improved resistivity uniformity: This is the core advantage of this method. It can reduce the ratio of the highest to the lowest resistivity of a single crystal rod (i.e., ρmax / ρ_min) to a very ideal range, far superior to the traditional CZ method. This is crucial for fields such as manufacturing high-voltage, high-power devices, where material uniformity is critical.

[0046] Improve material utilization and product yield:

[0047] In traditional methods, due to the large difference in resistivity between the head and tail, the portions of the crystal that exceed the specifications at both ends can only be cut off, resulting in material waste.

[0048] The resistivity of the single crystal rods produced by this method meets the standards within the effective length, and the usable length is greatly increased, which significantly improves the number of single crystals produced and the product yield.

[0049] High process controllability: By controlling the key parameter of "initial solidification ratio," the uniformity of the final resistivity can be flexibly adjusted. Process parameters can be customized according to different customers' requirements for resistivity range, realizing a shift from "passive acceptance" to "proactive design."

[0050] Reduced requirements for dopant control: Traditional methods require extremely precise calculations of the initial dopant dosage to obtain more uniform resistivity. However, this method, through a dynamic compensation system, has a certain "tolerance" and "correction" capability for small errors in the initial dopant, reducing the stringent requirements for initial process conditions. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the process steps of the present invention.

[0052] Figure 2 This is a schematic diagram of the solidification ratio versus resistivity curve of the present invention. Detailed Implementation

[0053] The technical solution of the invention will be further described in detail below through embodiments and in conjunction with the accompanying drawings.

[0054] Example: Figure 1 and Figure 2 As shown, a method for pulling silicon single crystals that reduces the resistivity range of the entire silicon single crystal rod includes the following steps:

[0055] Step 1: In crucible a, melt the solid silicon material and dopant together into a liquid.

[0056] Step 2: In crucible b, the crucible is slowly lowered to maintain the heating of the liquid in the upper part of the crucible, reduce the heating of the bottom of the crucible, and increase the heat dissipation from the bottom; thus, a portion of the liquid at the bottom of the crucible slowly solidifies into a solid.

[0057] Before starting the crystal pulling process, solid silicon is grown at the bottom of the crucible. Since the segregation coefficient of the dopant in silicon is less than 1, the concentration of the dopant in crucible b is increased relative to the concentration in crucible a, and the concentration of the dopant in the solid is decreased.

[0058] Before the pulling process begins, the dopant concentration is greater than that of the liquid in the crucible in stage a, while the dopant concentration in the solid is less than that of the liquid in the crucible in stage a. This is because the solid is formed gradually through solidification. Due to the segregation of dopant in silicon, the dopant concentration in the solid is not uniformly distributed, but the dopant concentration at each location is less than that of the liquid in the crucible in stage a.

[0059] By controlling the proportion of solid silicon to total silicon in crucible b, the ratio of the highest resistivity to the lowest resistivity of silicon single crystal rods can be reduced to varying degrees.

[0060] Figure 2 The horizontal axis represents the solidification ratio, which refers to the percentage of the weight of the grown solid silicon single crystal rod to the total amount of feed. Figure 2The vertical axis in the figure refers to the resistivity of the single-crystal silicon rod. When the initial feed amount and doping amount are the same, the resistivity distribution of the single-crystal silicon rod will be different when different pulling methods are used.

[0061] Step 3: After the crucible in crucible c is slowly lowered and stops, and the liquid to solid ratio in the crucible is stabilized, crystal pulling, shoulder formation, and equal diameter operation are performed.

[0062] Step 4: During the equal-diameter growth stage in crucible d, the heater gradually melts the solid at the bottom of the crucible into a liquid by raising the crucible level.

[0063] Step 5: The single crystal rod continues to grow in crucible e, and the liquid at the top of the crucible grows into a solid single crystal; at the same time, the solid silicon at the bottom of the crucible continues to melt into liquid from top to bottom, becoming a supplement to the liquid silicon material required for the growth of the single crystal rod at the top.

[0064] During the process in crucibles c, d, and e, the silicon single crystal at the top of the crucible gradually grows into a solid. Due to segregation, the concentration of dopant in the liquid increases. At the same time, the solid with a lower dopant concentration gradually melts into a liquid, which dilutes the dopant concentration in the liquid. This gradual dilution slows down the resistivity change of the single crystal rod along its length. The resistivity of the single crystal rod depends on the dopant concentration, thus reducing the resistivity range of the entire single crystal rod. This reduces the ratio of the highest resistivity to the lowest resistivity of the silicon single crystal rod when the solidification ratio is fixed.

[0065] Step 6: When the single crystal rod grows to the tail end in crucible f, all the solid silicon in the crucible melts into liquid, and the silicon single crystal pulling is completed through the tailing process.

[0066] Step 1 and Step 2: Establish an initial reserve of "high-concentration melt + low-concentration solid".

[0067] Objective: To prepare for subsequent "dilution".

[0068] Procedure: Before starting crystal pulling, allow a portion of the melt at the bottom of the crucible to solidify in a specific direction.

[0069] Principle: Since k < 1, the first solidified silicon will "repel" the dopant. Therefore, the dopant concentration Cs_solid of the solidified silicon at the bottom is much lower than the average concentration Cl_initial of the melt at this time. At the same time, because some dopant is "squeezed" into the remaining melt, the dopant concentration Cl_main of the main melt used for crystal pulling is significantly increased, higher than the average concentration at the initial feeding.

[0070] Key control point: By controlling the proportion of solidified solids at the bottom (i.e. Figure 2The "solidification ratio" setting allows for precise setting of the initial increase in Cl_main. A higher solidification ratio results in a higher initial concentration of Cl_main and a stronger subsequent compensation capability.

[0071] Steps 3 to 5: Dynamic crystal pulling and real-time compensation.

[0072] System state: At this point, a "liquid on top, solid on the bottom" coexistence system has been formed inside the crucible. The upper part is a highly doped melt, and the lower part is a low-doped solid.

[0073] Upper part (crystal pulling process): The single crystal grows from the upper melt. According to the segregation effect, the dopant concentration in the crystal, Cs_crystal = k * Cl_liquid. As the crystal grows, Cl_liquid should increase as in the conventional method.

[0074] Lower part (compensation process): By raising the crucible level, the heater gradually remelts the low-doped solid silicon at the bottom. The dopant concentration in this newly molten liquid is much lower than that in the upper main melt.

[0075] Core dynamic equilibrium: Dilution effect: The low-concentration "recycled liquid" continuously flows into the high-concentration "main melt", which plays a real-time dilution role on the dopant concentration Cl_liquid of the main melt.

[0076] Counteracting segregation: This dilution effect precisely offsets the upward trend of Cl_liquid caused by single crystal growth (segregation effect).

[0077] Stable concentration: By precisely controlling the melting rate of the bottom solid to match the growth rate of the upper crystal, the dopant concentration Cl_liquid of the main melt can be maintained at a relatively constant level throughout the constant diameter growth process.

[0078] Result: Since Cl_liquid is basically stable, the dopant concentration Cs_crystal=k*Cl_liquid of the grown single crystal also remains basically unchanged, thus achieving axial uniformity of resistivity of the entire single crystal rod.

[0079] Step 6: Finishing. As the single crystal growth nears completion, the compensating solid silicon at the bottom has completely melted and been consumed. The system returns to the traditional small-volume melt state for the final finishing operation. At this point, even if the resistivity changes, it only affects a very small portion at the end of the crystal, having a negligible impact on the uniformity of the entire crystal.

[0080] This study utilizes the segregation effect in semiconductor physics, transforming it from a "passive problem" into an "active control tool." When a material solidifies from a melt into a solid, the solubility of impurities (dopants) differs between the solid and liquid phases. This ratio is defined as the segregation coefficient: k = Cs / Cl, where Cs is the concentration of the impurity in the freshly solidified solid, and Cl is the concentration of the impurity in the melt.

[0081] In silicon: For the vast majority of dopants used as donors or acceptors in silicon (such as phosphorus P, boron B, arsenic As, etc.), their segregation coefficient k is less than 1.

[0082] The problem with the traditional Czochralski (CZ) method: In the traditional CZ method, as the single crystal grows, the amount of melt gradually decreases. Since k < 1, the dopant tends to remain in the melt, causing the dopant concentration (Cl) in the melt to continuously increase. Therefore, the resistivity of the first pulled crystal head (with a lower initial Cs value) is high, while the resistivity of the later pulled crystal tail (where Cs increases with increasing Cl) is low. This results in a significant difference in resistivity from beginning to end of the single crystal rod.

[0083] By establishing a dynamic system of "solidification at the top and melting at the bottom", a mechanism for "dilution" or "compensation" of the dopant concentration in the melt is created to counteract the unidirectional accumulation of Cl in traditional methods.

[0084] In summary, this silicon single crystal pulling method, which reduces the resistivity range of the entire silicon single crystal rod, effectively overcomes the inherent defect of axial resistivity non-uniformity caused by segregation effect in the traditional Czochralski method by constructing an initial state of "high-concentration melt + low-concentration solid" before crystal pulling and by controlling the melting rate of the bottom solid during crystal pulling to dilute and compensate the dopant concentration of the main melt in real time and dynamically.

[0085] The above description is only a specific embodiment of the present invention, but the structural features of the present invention are not limited thereto. Any changes or modifications made by those skilled in the art within the scope of the present invention are covered by the patent scope of the present invention.

Claims

1. A method for pulling silicon single crystals that reduces the resistivity range of the entire silicon single crystal rod, characterized in that... The following steps are included: Step 1: In crucible a, melt the solid silicon material and dopant together into a liquid; Step 2: In crucible b, the crucible is slowly lowered to maintain the heating of the liquid in the upper part of the crucible, reduce the heating of the bottom of the crucible, and increase the heat dissipation from the bottom; so that a part of the liquid at the bottom of the crucible slowly solidifies into a solid. Step 3: After the crucible in crucible c is slowly lowered and stops, and the liquid to solid ratio in the crucible is stabilized, crystal pulling, shoulder formation and equal diameter operation are performed. Step 4: During the equal-diameter growth stage in crucible d, the heater gradually melts the solid at the bottom of the crucible into a liquid by raising the crucible level. Step 5: The single crystal rod continues to grow in crucible e, and the liquid at the top of the crucible grows into a solid single crystal; at the same time, the solid silicon at the bottom of the crucible continues to melt into liquid from top to bottom, becoming a supplement to the liquid silicon material required for the growth of the single crystal rod at the top. Step 6: When the single crystal rod grows to the tail end in crucible f, all the solid silicon in the crucible melts into liquid, and the silicon single crystal pulling is completed through the tailing process.

2. The method for pulling silicon single crystals according to claim 1, which reduces the resistivity range of the entire silicon single crystal rod, is characterized in that: Before starting the crystal pulling process, solid silicon is grown at the bottom of the crucible. Since the segregation coefficient of the dopant in silicon is less than 1, the concentration of the dopant in crucible b is increased relative to the concentration in crucible a, and the concentration of the dopant in the solid is decreased.

3. The method for pulling silicon single crystals according to claim 2, which reduces the resistivity range of the entire silicon single crystal rod, is characterized in that: Before the pulling process begins, the dopant concentration is greater than that of the liquid in the crucible in stage a, while the dopant concentration in the solid is less than that of the liquid in the crucible in stage a. This is because the solid is formed gradually through solidification. Due to the segregation of dopant in silicon, the dopant concentration in the solid is not uniformly distributed, but the dopant concentration at each location is less than that of the liquid in the crucible in stage a.

4. The method for pulling silicon single crystals according to claim 1, which reduces the resistivity range of the entire silicon single crystal rod, is characterized in that: During the process in crucibles c, d, and e, the silicon single crystal at the top of the crucible gradually grows into a solid. Due to segregation, the concentration of dopant in the liquid increases. At the same time, the solid with a lower dopant concentration gradually melts into a liquid, which dilutes the dopant concentration in the liquid. This gradual dilution slows down the resistivity change of the single crystal rod along its length. The resistivity of the single crystal rod depends on the dopant concentration, thus reducing the resistivity range of the entire single crystal rod. This reduces the ratio of the highest resistivity to the lowest resistivity of the silicon single crystal rod when the solidification ratio is fixed.

5. The method for pulling silicon single crystals according to claim 1, characterized in that: By controlling the proportion of solid silicon to total silicon in crucible b, the ratio of the highest resistivity to the lowest resistivity of silicon single crystal rods can be reduced to varying degrees.

6. The method for pulling silicon single crystals according to claim 5, which reduces the resistivity range of the entire silicon single crystal rod, is characterized in that: The horizontal axis represents the solidification ratio, which refers to the percentage of the weight of the grown solid silicon single crystal rod to the total feed amount; the vertical axis represents the resistivity of the silicon single crystal rod. When the initial feed amount and dopant amount are the same, the resistivity distribution of the silicon single crystal rod is different when different pulling methods are used.

Images