TWI311168B - Method for manufacturing silicon monocrystal - Google Patents

Method for manufacturing silicon monocrystal Download PDF

Info

Publication number
TWI311168B
TWI311168B TW92126275A TW92126275A TWI311168B TW I311168 B TWI311168 B TW I311168B TW 92126275 A TW92126275 A TW 92126275A TW 92126275 A TW92126275 A TW 92126275A TW I311168 B TWI311168 B TW I311168B
Authority
TW
Taiwan
Prior art keywords
single crystal
crystal rod
diameter
rod
pulling
Prior art date
Application number
TW92126275A
Other languages
Chinese (zh)
Other versions
TW200512321A (en
Inventor
Daisuke Wakabayashi
Masao Saito
Satoshi Sato
Jun Furukawa
Kounosuke Kitamura
Original Assignee
Sumitomo Mitsubishi Silicon
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Sumitomo Mitsubishi Silicon filed Critical Sumitomo Mitsubishi Silicon
Priority to TW92126275A priority Critical patent/TWI311168B/en
Publication of TW200512321A publication Critical patent/TW200512321A/en
Application granted granted Critical
Publication of TWI311168B publication Critical patent/TWI311168B/en

Links

Landscapes

  • Liquid Deposition Of Substances Of Which Semiconductor Devices Are Composed (AREA)
  • Crystals, And After-Treatments Of Crystals (AREA)

Description

1311168 玖、發明說明: 【發明所屬之技術領域】 本發明係關於一種藉PID控制法(溫度控制法)控制以 切克勞斯基法(czoch-raski methed)拉起之石夕單結晶棒之直 徑之矽單結晶品之製造方法。 【先前技術】 以往,石夕單結晶之拉晶速度的控制方法已被人所知者 有例如在矽單結晶棒之提拉中,將直徑之偏差直接回饋 (feed back)至單結晶棒之拉晶速度之第1方法,及將上述之 直徑偏差直接回饋至加熱器溫度(heater temperature)之第2 方法。 由於近年隨著半導體元件之高積集化,設計規則 (design rule)更為微細化,使材料之晶圓上之微細缺陷大大 的影響元件之收率。因此沿提拉之單結晶棒之軸正交(直角) 面切割製造晶圓時’需要製造晶圓的全面無微細缺陷之晶 圓。為此目的,設單結晶棒之提拉時之固液界面近旁之軸 方向的溫度梯度為G(°C /mm),其提拉速度為V(mm/分)時, 需將提拉速度全程設定成一定V/G並控制於該設定之提拉 速度。保持V/G全程一定時,由於單結晶棒之提拉初期之 頂部溫度梯度G大,自頂部至所定之提拉位置溫度梯度變 小,故配合上述溫度梯度G的變化設定提拉速度,通常需 要將頂部之提拉速度定為較快,並次第的減慢至所定之提 拉位置之提拉速度。 但,此種設定之提拉速度之差異將直接變為實際之提 1311168 =之,異’故依上述之第1方法提高頂部之控制性, =以外之直徑變動會增大,想提高頂部以外之控制性 則又有頂部之直徑變動增大之問題發生。 &上述之第2方法時,由於加熱器之控制幅度及 :液:度之變化時間會隨液面與加熱器之位置關係或融液 化使’里度之控制變為很困難,有時甚至會導致實 際之提拉(拉晶)速度之變化方向與加熱器之補正方向變為 不一致,使直徑變動增大之虞。 —為解決k二問題,在日本特開細U⑹叩號公報有揭 種運算石夕單結曰曰棒之提拉速度之控制值,對此提拉速 度,控制值進行提拉速度之幅度調整(span __,並 對”亥運算之提拉速度之控制值進行幅度調整前,經由提拉 速度^控制與設定之提拉速度之比較,運算加熱器溫度之 補正量求得加熱器溫度之設定輸出而控制矽單結晶棒之直 徑為特徵之矽單結晶之製造方法。 日但、’上述公報揭示之石夕單結晶之製造方法,由於是在 對提拉速度之控制值進行幅度調整前,比較提拉速度控制 值與設定之提拉速度而將其偏差回饋至加熱器溫度,故有 t熱器溫度之補正量不能追隨實際之提拉速度之偏差且有 單結晶棒之直徑變動增大之問題。 ,本=明之目的為提供-種可將石夕單結晶棒之設定提拉 速度认疋為-定之V/G ’根據此設定提拉速度精確地控制 實際之提拉速度,從而抑制單結晶棒之直徑變動為特徵之 矽單結晶之製造方法。 1311168 本發明之另一目的為媒很——p J_ , 的為如供種可在矽單結晶棒之提拉 時預測有無發生不良部,藉修& D。1危 精乜正5又疋楗拉速度及設定加熱 益棚度,從而減少或阻止矽單 枯佩A y早、、·。曰日棒中之不良部之產生為 特徵之矽皁結晶之製造方法。 〃申請專利範圍第i項之發明為,如第1及2圖所示, 係從加熱器17所融解之㈣液13提拉♦單結晶棒24,依 所定之時間間隔檢測該提拉中之石夕單結晶棒2 4之直徑變 化’將此檢測值回饋(feedback)至石夕單結晶棒24之提拉速 ,及加熱器17溫度,從而控制矽單結晶棒24之直徑之矽 單結晶製造方法之改良。 其構成之特徵包括: 胃使石夕單結晶棒24的直徑達到目標值之石夕單結晶棒24 提拉速度之控制方法;控制加熱器17溫度,使梦單結晶棒 的直‘達到目私值之加熱器溫度控制方法;以及在該等 方法之複數階段實行適用改變PID常數之pm控制。 依此申請專利範圍第i項之石夕單結晶之製造方法是, 石夕單結晶棒24之頂部/頭部(t〇p/head)之提拉時,設定提拉 速度快(高),其直徑變動亦有變大之傾向,為使此頂部之直 徑早期安定,藉將PID控制之PID常數設定較大,以增大 對矽單結晶棒24之直徑偏差之提拉速度之補正量,從而優 先實仃控制矽單結晶棒24之直徑控制。又,矽單結晶棒24 之頂部以後(即直軀部之底部)之提拉時,設定提拉速度較頂 部為慢(低),其直徑變動亦有變小之傾向,於是因應設定提 拉速度,將PID控制之PID常數階段的變小,以減小對石夕 1311168 早、,,σ日日棒24之直徑偏差之提拉速度之補正量,4 + 行控财單結晶棒24之提拉,二:正“而優先貫 階段之PID㈣ ^拉速度之控制。在此所謂「複數 工1」如上述係指使PID控制之PID常數階段 的變化之意。 ^数丨白丰又 申請專利範圍第2項之發明為, 伟於制石夕星社曰a 知月為如第1及2圖所示, ^财早結晶棒24之提拉速度使其直徑達到目標值之 PID控制方法中將石夕單έ士曰基 直泸傯#吉姑 早…曰棒24之目標直徑及實測直徑之 二偏差直接回饋之方法及將上述直徑偏差 偏差回饋至現在之趄里1 f马 在之&拉錢之回饋方法的組合為特徵。 社曰棒24 ^所記载之料結晶之製造方法係在提拉石夕單 二曰二Ϊ 後之提拉時’以對前^之提拉速度作 ρ ID控制回饋時之石夕單結晶棒2 又作 直徑偏差作為基準,而^ Α目標直徑及實測直徑之 實測直和之M結晶棒之目標直捏及 速声,心:^ 化量作為偏差’補正前次之提拉 棒二之提^—/卩㈣部㈣之提料發生W單結晶 释24之扼拉速度之變動。 件以^項之發明為’如第1及2圖所示, 變化旦你:之目標直徑及實測直徑之直徑偏差之 偏差㈣切單結晶棒24之提拉速度時,將上 速f加以PiD控制,使其不超過對現在之提拉速卢 之補正之联大變動幅度為特徵。 逮a 棲24依Ϊ第3項記载之石夕單結晶之製造方法,於石夕單-曰 棒%之提拉時,以前次之石夕 夕::: 基準,將石夕單結晶棒24之直+之&拉速度作為 之直&偏差之變化量作為偏差回饋 1311168 至此次之矽單結晶棒24之提拉速度時,若是超過對現在之 提拉速度之補正之最大變動幅度,則此補正會受最大變動 幅度之限制,於是得以將矽單結晶棒24之提拉速度之變動 抑制於最小限。 申請專利範圍第4項之發明係根據第1〜3項之發明, 由第3〜5及11圖進一步題示,係以矽單結晶棒24之提拉 時,使用提拉開始至所定時刻止之提拉速度實測設定檔 (profile)及提拉開始至提拉終了止之設定提拉速度,併行矽 單結晶棒之品質預測運算;當預測會有矽單結晶棒24之不 良部發生,則運算修正不良部所要之矽單結晶棒24之修正 提拉速度及修正加熱器溫度而將此等修正提拉速度及修正 加熱器溫度回饋至設定提拉速度及設定加熱器溫度為特 徵。 依此第4項記載之矽單結晶之製造方法,於矽單結晶 棒24之提拉時,為保持所定直徑,即使實測提拉速度超過 設定提拉速度之範圍之場合,由於是從提拉速度之實績及 設定逐次併行矽單結晶棒之品質預測運算,故可逐次預測 有無不良部之發生。結果,可算去用以修正該不良部之矽 單結晶棒24之修正提拉速度及修正加熱器溫度並將此等修 正提拉速度及修正加熱器溫度回饋,從而減低或阻止不良 部之發生。 申請專利範圍第5項之發明係根據第4項之發明,由 第6〜12圖顯示,其矽單結晶棒24之品質之預測計算及修 正提拉速度之求算係藉使用含有: 10 1311168 η二=單結晶棒24之製造條件之參數Pl,考慮石夕融液 出切㈤液13成長之⑪單結晶棒24内之溫度 分佈之求算步驟; t =求出冷卻過程之石夕單結晶棒24内之溫度分佈預 ::早、w曰棒内之孔隙(void)及高溫氧析出物之濃度分佈 及大小(size)分佈之步驟; ⑺於計算求㈣單結濃度線及第】 2線t後,藉計算求出第1等濃度線之變曲點之最大值 一第1分佈線之變曲點之最小值之差之步驟; ⑷將矽單結晶棒24之製造條件之參數,’由p2依序改 變至PN而計算求出第1等濃度線之變曲點之最大值與第i 刀佈值之變曲點之最小值之差之步驟;及 ⑺求出可使第1等濃度線之變曲點之最大值與第}分 佈線之變曲點之最小值之差為最大之石夕單結晶棒24之製造 條件之步驟;之電腦而使料結晶棒24之無缺陷領域最大 化之缺陷模擬(仿真)方法(defect simulad〇n _〇句實施為 特徵。 依上述申請專到範圍帛5項之石夕單結晶棒之製造方法 時,不僅可求出由石夕單結晶棒24之製造條件之參數A,考 虑石夕融液13之對流求出由石夕融液13成長之石夕單社 之溫度分佈’ ^可求出冷卻過程之石夕單結晶棒 分佈,即考慮由矽融液拉出之矽單結晶棒24之冷卻過程之 石夕單、’、。BB棒之知冷及急冷效果加以解析;預測石夕單尹曰棒 内之空隙(v。i d)之濃度分佈及大小分佈並同時預測石夕y曰结 1311168 晶棒内之高溫氧析出物之濃度分佈及大小分佈。繼之,通 過計算求出矽單結晶棒内之第j等濃度線Hcix及第i分佈 線BClx之後,再通過計算求出第1等濃度線之變化 點之最大值與第1分佈線BC;U之變化點之最小值之差△ Z!。 後將上述石夕單結晶棒之製造條件之參數,由p2依序 改變至fN’並與計算求出第1等濃度值HClx之變化點之最 大值與第1分佈值BCU變曲點之最小值之差而後 進了步計算求出上述第1等濃度ft HCUt曲點之最大值 ,第1分料BCU變曲點之最小值之差ΔΖι會最大之石夕 單、’、。aa棒之製造條件。依此便可通過計算正確的求出矽單 ,晶棒24之提拉方向及半徑方向之無缺陷領域最擴大之矽 單結晶棒24之製造條件。 申請專利範圍第6項之發明係根據第4項之發明,由 =丨2圖顯不,其矽單結晶棒24之品質之預測計算及修正 =拉速度之運算係以包含:(1)於矽單結晶棒24之製造條件 =數Pi,考慮融液13之對流求出由矽融液13成長之矽單 f晶棒24内之溫度分佈之步驟;(2)求算冷卻過程之矽單結 二棒内之溫度分佈,從而預測矽單結晶棒内之空隙及 门'皿氧析出物之濃度分佈及大小分佈之步驟;(3)於計算求 $矽單結晶棒内之第2等濃度線及第2分佈線之後,再計 算长出第2等濃度線之變曲點之最大值與第2分佈線之變 ‘‘之敢小值之差之步驟;(4)將;5夕單結晶棒之製造條件之 參數,從1>2依序改變至?>1而求算第2等濃度線之變曲點之 12 1311168 最大值及第2分佈線之變曲點之最小值之差之步驟;及 求算第2等遭度線之變曲點之最大值與第2分佈線之變曲 點之最小值之差會最大之單結晶棒之製造條件之步驟,為 其特徵之一種藉使用電腦將矽單結晶棒之無女 化之缺陷模擬方法實施。 ㈣最大 依此申請專利範圍第6項之矽單結晶之製造方法時, 於石夕單結晶棒24之製造條件參數P丨與申請專利範圍第5 項同樣預測石夕單結晶棒之空隙之濃度分佈及大小分佈,同 時預測發單結晶棒内之高溫氧析出物及濃度及大小分佈。 繼之,通料算求出料結晶棒内之第2等濃度線HC2x及1311168 发明, invention description: [Technical field to which the invention pertains] The present invention relates to a method for controlling a stone singular crystal rod which is pulled up by the czoch-raski methed method by a PID control method (temperature control method) A method for producing a single crystal product of a diameter. [Prior Art] In the past, a method for controlling the pulling speed of the singular single crystal has been known, for example, in the pulling of a single crystal rod, and the deviation of the diameter is directly fed back to the single crystal rod. The first method of pulling the crystal speed and the second method of directly feeding back the above-described diameter deviation to the heater temperature. In recent years, with the high integration of semiconductor components, the design rule has been further refined, so that the micro defects on the wafer of the material greatly affect the yield of the device. Therefore, when the wafer is manufactured by orthogonal (right-angle) plane cutting along the axis of the single crystal rod of the pulling, it is necessary to manufacture a wafer with a fine defect-free crystal. For this purpose, the temperature gradient in the axial direction near the solid-liquid interface at the time of pulling the single crystal rod is G (°C / mm), and the pulling speed is V (mm/min), and the pulling speed is required. The whole process is set to a certain V/G and controlled to the set pulling speed. When the V/G is maintained for a certain period of time, since the top temperature gradient G at the initial stage of pulling of the single crystal rod is large, the temperature gradient from the top to the predetermined pulling position becomes small, so the pulling speed is set in accordance with the change of the temperature gradient G, usually It is necessary to set the lifting speed of the top to be faster, and to slow down to the pulling speed of the predetermined pulling position. However, the difference in the pulling speed of such a setting will directly become the actual increase of 1311168 =, and therefore, according to the first method described above, the controllability of the top is improved, and the diameter variation other than = is increased, and it is desired to increase the The controllability has the problem of an increase in the diameter of the top. & The second method described above, because the control range of the heater and the change time of the liquid: degree will become difficult with the positional relationship between the liquid surface and the heater or the liquefaction, so that even the control of the degree is difficult, sometimes even This will cause the actual direction of the pulling (drawing) speed to change to the direction of the heater's correction, which will increase the diameter variation. - In order to solve the problem of k2, in the Japanese special open U(6) nickname bulletin, the control value of the lifting speed of the Shixi single knot crowbar is revealed, and the pulling speed and the control value are adjusted for the lifting speed. (span __, and before the amplitude adjustment of the control value of the lifting speed of the hai operation, the comparison between the lifting speed control and the set pulling speed is performed, and the correction of the heater temperature is calculated to obtain the heater temperature setting. The method for producing a single crystal which is characterized by controlling the diameter of the single crystal rod by the output. The method for producing the single crystal of the stone disclosed in the above publication is because the amplitude of the control value of the pulling speed is adjusted. Comparing the pulling speed control value and the set pulling speed and feeding the deviation back to the heater temperature, the correction amount of the t-heater temperature cannot follow the deviation of the actual pulling speed and the diameter variation of the single crystal rod increases. The problem is that the purpose of this = Ming is to provide - the type of pull-up speed of the Shixi single crystal rod can be regarded as - set V / G 'According to this set pull speed to accurately control the actual pull speed Therefore, a method for producing a single crystal which is characterized by variation in the diameter of a single crystal rod is suppressed. 1311168 Another object of the present invention is that the medium is very large, p J_ , which is such that the seed can be predicted during the pulling of the single crystal rod. The occurrence of a bad department, borrowed & D. 1 dangerous fine 乜 疋楗 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 The invention relates to a method for producing cerium soap crystals characterized by the invention. The invention of claim i is as shown in Figs. 1 and 2, which is a liquid crystal 13 which is melted by the heater 17 and is pulled by a single crystal rod 24 Detecting, according to the determined time interval, the diameter change of the single crystal rod 24 in the pulling, 'feeding the detection value to the pulling speed of the stone single crystal rod 24, and the temperature of the heater 17, thereby The invention relates to an improvement of the method for manufacturing a single crystal of the diameter of the single crystal rod 24. The characteristic of the composition includes: a method for controlling the pulling speed of the stone single crystal rod 24 whose diameter reaches the target value by the stomach; Control the temperature of the heater 17 to make the dream single crystal rod straight' a heater temperature control method to a private value; and a pm control adapted to change the PID constant at a plurality of stages of the methods. The manufacturing method of the stone single crystal according to the item i of the patent application is: When pulling the top/head of the rod 24 (t〇p/head), the pulling speed is set to be fast (high), and the diameter variation also tends to become larger. In order to stabilize the diameter of the top at the beginning, the PID is borrowed. The PID constant of the control is set to be large to increase the correction amount of the pulling speed of the diameter deviation of the single crystal rod 24, thereby preferentially controlling the diameter control of the single crystal rod 24. Further, the single crystal rod 24 is When pulling at the top (ie, the bottom of the straight body), the pulling speed is set to be slower (lower) than the top, and the diameter variation also tends to become smaller. Therefore, the PID constant of the PID control is set in response to the setting of the pulling speed. The stage becomes smaller, so as to reduce the correction amount of the pulling speed of the diameter deviation of the shoal 1311168 early, and σ day bar 24, the lifting of the 4 + line control financial crystal rod 24, two: " PID (four) of the priority phase ^ control of the pull speed. Here, "plural worker 1" as described above means the change of the PID constant stage of the PID control. ^The number of inventions in which Baifeng has applied for the second paragraph of the patent scope is: Wei Yushi Shixing Xingshe a Zhiyue is as shown in Figures 1 and 2, the lifting speed of the early crystallization rod 24 makes its diameter In the PID control method that achieves the target value, the method of directly returning the deviation of the target diameter and the measured diameter of the target diameter of the 吉 早 έ 曰 泸偬 吉 吉 吉 吉 及 及 及 及 及 及 及 及 及 及 24 24 24 24 24 24 24 24趄里1 f马 is characterized by a combination of the < The manufacturing method of the crystallization of the material described in the 曰 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 Rod 2 is also used as the reference for the diameter deviation, and ^ Α target diameter and measured diameter of the measured direct and the M crystal rod target straight pinch and speed sound, heart: ^ amount as a deviation 'correction before the lifting rod two Lifting of the ^-/卩(4) part (4) occurs when the pulling speed of the W single crystal release 24 occurs. The invention of the item is 'as shown in Figures 1 and 2, the variation is the deviation of the diameter deviation between the target diameter and the measured diameter. (4) When the pulling speed of the single crystal rod 24 is cut, the upper speed f is added to the PiD. Control so that it does not exceed the change in the General Assembly's corrections to the current lifting speed. The method of manufacturing the stone singular crystals recorded in the third item of a habitat 24, in the case of the Shi Xidan-曰 % % , , 以前 以前 以前 以前 以前 以前 以前 以前 以前 以前 以前 以前 以前 以前 以前 以前 以前 : : : : : : : : : : The direct change of the straightness & Then, the correction is limited by the maximum fluctuation range, so that the variation of the pulling speed of the single crystal rod 24 can be suppressed to a minimum. The invention of claim 4 is based on the inventions of the first to third aspects, and is further described by the third to fifth and eleventh figures, when the pulling of the single crystal rod 24 is performed, the pulling is started until the predetermined time. The pulling speed measurement profile (profile) and the lifting speed from the start of the lifting to the end of the lifting, and the quality prediction operation of the single crystal rod in parallel; when it is predicted that the defective portion of the single crystal rod 24 occurs, The correction pulling speed and the corrected heater temperature of the single crystal rod 24 required by the correction correction unit are characterized by the correction of the pulling speed and the correction of the heater temperature to the set pulling speed and the setting of the heater temperature. According to the manufacturing method of the single crystal according to the fourth item, when the single crystal rod 24 is pulled, the diameter is maintained, and even if the measured pulling speed exceeds the set pulling speed, the pulling is performed. The actual performance of the speed and the setting of the quality prediction calculation of the single crystal rod in parallel are successively predicted, so that the occurrence of the defective part can be predicted one by one. As a result, it is possible to correct the correction pulling speed of the single crystal rod 24 of the defective portion and correct the heater temperature, and to correct these lifting speeds and correct the heater temperature feedback, thereby reducing or preventing the occurrence of defective portions. According to the invention of claim 4, the invention according to the fourth aspect shows that the calculation of the quality of the single crystal rod 24 and the calculation of the modified pulling speed are used as follows: 10 1311168 η 二 = parameter P1 of the manufacturing conditions of the single crystal rod 24, taking into account the calculation step of the temperature distribution in the 11 single crystal rod 24 in which the Shi Xirong liquid cuts out (5) liquid 13 grows; t = finds the cooling process The temperature distribution in the crystal rod 24 is as follows: the steps of the concentration distribution and size distribution of the pores and the high temperature oxygen precipitates in the early, w曰 rods; (7) the calculation of the (four) single knot concentration line and the first] After the 2nd line t, the step of calculating the difference between the maximum value of the inflection point of the first equal concentration line and the minimum value of the inflection point of the first distribution line is calculated by calculation; (4) The parameters of the manufacturing conditions of the single crystal rod 24 , 'the step of calculating the difference between the maximum value of the inflection point of the first iso-concentration line and the minimum value of the inflection point of the i-th knife cloth value by changing p2 to PN sequentially; and (7) finding the same The difference between the maximum value of the inflection point of the 1 equal concentration line and the minimum value of the inflection point of the distribution line is the largest The step of manufacturing the ingot 24; the defect simulation (simulation) method for maximizing the defect-free area of the crystallizing rod 24 by the computer (defect simulad〇n _ 〇 实施 is implemented as a feature. In the method for producing the singular single crystal rod, the parameter A of the manufacturing conditions of the shixi single crystal rod 24 can be obtained, and the convection of Shi Xirong liquid 13 can be used to determine the eve of the growth of Shi Xirong liquid 13 The temperature distribution of the single society can be used to determine the distribution of the single crystal rod of the cooling process, that is, the cooling process of the single crystal rod 24 pulled out by the mash melt, and the BB rod is cold. And the quenching effect is analyzed; the concentration distribution and size distribution of the void (v. id) in the Shixidan Yinyu rod are predicted and the concentration distribution and size distribution of the high temperature oxygen precipitates in the 1311168 ingot of Shixi y曰jie are predicted at the same time. Then, after calculating the j-th equal concentration line Hcix and the i-th distribution line BClx in the single crystal rod, the maximum value of the change point of the first iso-concentration line and the first distribution line BC are calculated by calculation. The difference between the minimum points of the change points of U △ Z!. The parameters of the manufacturing conditions of the above-mentioned stone-like single crystal rod are sequentially changed from p2 to fN' and the maximum value of the change point of the first equal-concentration value HClx and the minimum value of the BCU inflection point of the first distribution value are calculated. After the difference is calculated, the maximum value of the first equal-concentration ft HCUt curve is obtained, and the difference ΔΖι of the minimum value of the BCU inflection point of the first material is the largest. According to this, the manufacturing conditions for the single-crystal rod 24 which is the most enlarged in the non-defective field of the pulling direction and the radial direction of the ingot 24 can be obtained by calculation. According to the invention of the fourth aspect, the calculation of the quality of the single crystal rod 24 and the calculation of the correction = the pulling speed are as follows: (1) The manufacturing conditions of the single crystal rod 24 are: a number Pi, a step of determining the temperature distribution in the single f-crystal rod 24 grown by the crucible melt 13 in consideration of the convection of the melt 13; (2) calculating the temperature distribution in the single-junction two rods of the cooling process, thereby Predicting the distribution and size distribution of the voids in the single crystal rod and the oxygen precipitates in the door Step (3) After calculating the second equal concentration line and the second distribution line in the single crystal rod, calculate the maximum value of the inflection point of the second equal concentration line and the change of the second distribution line. ''The step of the difference between the small value of the dignity; (4) will; the parameters of the manufacturing conditions of the single crystal rod of 5 eve, change from 1 > 2 to ? >1, the step of calculating the difference between the maximum value of the 12 1311168 of the inflection point of the second equal concentration line and the minimum value of the inflection point of the second distribution line; and calculating the inflection point of the second equalization line The step of manufacturing the condition of the single crystal rod having the largest difference between the maximum value of the second distribution line and the minimum value of the inflection point of the second distribution line, and a characteristic of the method for simulating the defect of the monocrystalline rod by using a computer Implementation. (4) When the manufacturing method of the single crystal of the sixth paragraph of the patent application scope is the largest, the manufacturing condition parameter P丨 of the Shixi single crystal rod 24 is the same as the fifth paragraph of the patent application scope, and the concentration of the void of the stone single crystal rod is predicted. Distribution and size distribution, and predict the high temperature oxygen precipitates and concentration and size distribution in the crystallized rod. Then, the second equal concentration line HC2x in the crystal rod of the material is calculated and calculated.

第2分佈線BC2x之後,再通過計算求出帛2等濃度線⑽X 之變曲點之最大值與第2分佈線Β(:2χ之變曲點之最小值之x 差△ Z2。 爾後’將上料單結晶棒之製造條件之參數,從匕依 序改變SPN ’並與上述同樣,通過計算求出第2等濃度線 HC2X之變曲點之最大值與第2分佈線之變曲點之最小 ^之差ΔΖ”進而求出域第2等濃度線HC2X之變曲點之 最大值與^ 2分佈線响之變曲點之最小值之差△心。會 最大之Γ單結晶棒之製造條件。依此便可通過計算正確的 2石夕單結晶棒24之提拉方向及半徑方向之無缺陷領域最 擴大之矽單結晶棒之製造條件。 【實施方式】 羡依圖示說明本發明之第1實施形態於下。. 如第1圖所示’石夕單結晶之提拉裝置1〇,即拉晶裝置 13 1311168 備有内令可呈真空之主室(main chamber)ll及裝設於該室 内中央之坩堝12。該主室11為圓筒狀之真空容器。坩堝 係由石英製成之存放矽融液13用有底圓筒狀内層容器 仏及—由。石墨製成之外嵌於上述内層容器12a之有底圓筒狀 外層今益12b所構成。外層容器12b之底面係連接於坩堝 軸14之上端,而此坩堝軸14之下端設有一使坩堝轉動且 昇降之驅動裝置16。另外,在_12之外周面隔著一定之 間:宋包,狀的設有加熱器17,同時該加熱器17之外周面則 隔著疋之間隙包圍狀的設有圓筒狀之保溫筒〗8。 又,主室11之上端連接有一直徑小於主室直徑之圓筒 狀拉晶室19 °此拉晶冑19之上端設有拉晶裝置(未圖示), 而該拉晶裝置具有可供其下端抵達主室11内之石夕融液13 ^面之拉晶轴旋轉及昇降之構造。上述拉晶軸21之下 端設有種晶夾頭(seed chuck;) 23 可藉此夾頭夾住種晶22。當將該種曰曰曰22之下端浸泡於石夕融 ,13中後,藉拉晶裝置使種晶22及㈣22分別回轉及提 幵’即可從種晶22之下端拉起成長之梦單結晶棒24。 上述拉起中之固液界面近旁之石夕單結晶棒24之直徑係 用直担檢測感應器(未圖示)依所定時間間隔檢測。此直徑檢 ,感應㈣由電荷麵合元件相機(CCDCamera)及放射溫度 "一十等構成。直徑檢測感應器之檢測輸出接續於控制器(未圖 :)之控制輸入’而控制器之控制輸出則接續於加熱器Η, =晶襄置及_驅純置16。控制器設有記憶體,此記憶 體儲存有提拉之料結晶棒24之目標直徑、v/g全程設定 14 1311168 為一定之設定提拉速度及V/G全程設定為一定之設定加熱 器溫度為資料作為映射圖像(map)加以記憶,其中V為矽單 結晶棒之提拉速度(拉晶速度)、G為矽單結晶棒之提拉時之 固液界面近旁之軸方向的溫度梯度。 控制器根據直徑檢測感應器之檢測輸入控制加熱器 17、拉晶裝置及坩堝驅動裝置16。即,此控制器如第1及 2圖所示,將上述直徑變化之檢測輸出回讀至加熱器1 7、 拉晶裝置及坩堝驅動裝置16進行矽單結晶棒24之直徑控 制。矽單結晶棒24之直徑控制方法可使用(1)優先對矽單結 晶棒24之提拉速度進行PID控制,使矽單結晶棒之直徑達 到目標值之方法,及(2)優先對加熱器17溫度進行PID控 制,使矽單結晶棒之直徑達到目標值之方法。在此等方法 中可適當採用多階段之改變PID常數之PID控制。PID控 制為回饋控制之一方式,係將比例於系之輸出之信號、系 之輸出積分信號及系之輸出微分信號合併使用之控制方 式。 具體之方法乃依下式(A)控制提拉(拉晶)速度,使控制 器在矽單結晶棒24之頂部提拉時優先進行直徑控制,在提 拉頂部以後之部份時優先進行提拉速度控制:After the second distribution line BC2x, the maximum value of the inflection point of the concentration line (10)X such as 帛2 is calculated and the x difference ΔZ2 of the minimum value of the second distribution line Β (the minimum of the inflection point of 2χ). The parameters of the manufacturing conditions of the single crystal rod are changed, and the SPN' is changed sequentially from the crucible, and the maximum value of the inflection point of the second iso-concentration line HC2X and the inflection point of the second distribution line are calculated by the same calculation as described above. The minimum difference ΔΖ” further determines the difference between the maximum value of the inflection point of the second equal concentration line HC2X of the domain and the minimum value of the inflection point of the distribution line of the ^2. The largest single crystal rod is produced. According to this, it is possible to calculate the manufacturing conditions of the most expanded single crystal rod in the non-defective field in the pulling direction and the radial direction of the correct 2 stone single crystal rod 24. [Embodiment] The first embodiment is as follows: As shown in Fig. 1, the pull-up device 1 of the lithium single crystal, that is, the crystal pulling device 13 1311168, is provided with a main chamber ll which is capable of being vacuumed and loaded. The crucible 12 is disposed in the center of the room. The main chamber 11 is a cylindrical vacuum container. The prepared storage crucible 13 is composed of a bottomed cylindrical inner layer container and a bottomed cylindrical outer layer of the outer layer container 12a, which is formed of graphite. It is connected to the upper end of the boring shaft 14, and the lower end of the boring shaft 14 is provided with a driving device 16 for rotating and lifting the raft. In addition, the outer surface of the _12 is separated by a certain distance: At the same time, the outer peripheral surface of the heater 17 is surrounded by a gap of a crucible, and is provided with a cylindrical heat insulating tube 8. The upper end of the main chamber 11 is connected with a cylindrical pull having a diameter smaller than the diameter of the main chamber. The crystal chamber 19 ° is provided with a crystal pulling device (not shown) at the upper end of the crystal pulling crucible 19, and the crystal pulling device has a crystal pulling shaft rotation for the 13th surface of the Shixi melt which can reach the lower end of the chamber 11 And the structure of lifting and lowering. The lower end of the crystal pulling shaft 21 is provided with a seed chuck; 23, the seed crystal 22 can be clamped by the chuck. When the lower end of the seed 22 is soaked in Shi Xirong After 13th, the seed crystal 22 and (4) 22 are respectively rotated and lifted by the crystal pulling device, and the lower end of the seed crystal 22 can be obtained. The growth dream single crystal rod 24. The diameter of the Shixi single crystal rod 24 in the vicinity of the solid-liquid interface in the above-mentioned pull-up is detected by a straight-line detecting sensor (not shown) at a predetermined time interval. (4) It consists of a CCD camera and a radiation temperature. The detection output of the diameter detection sensor is connected to the control input of the controller (not shown) and the control output of the controller is connected to the heating. Η, = 襄 襄 _ _ 纯 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器 控制器The speed and V/G are set to a certain value. The heater temperature is stored as a map image, where V is the pulling speed of the single crystal rod (drawing speed) and G is the single crystal rod. The temperature gradient in the axial direction near the solid-liquid interface at the time of pulling. The controller controls the heater 17, the crystal pulling device, and the crucible driving device 16 based on the detection input of the diameter detecting sensor. That is, as shown in Figs. 1 and 2, the controller returns the detection output of the diameter change to the heater 17 and the crystal pulling device and the crucible driving device 16 to control the diameter of the single crystal rod 24. The method for controlling the diameter of the single crystal rod 24 can be used by (1) preferentially controlling the pulling speed of the single crystal rod 24, controlling the diameter of the single crystal rod to reach the target value, and (2) preferentially heating the heater. 17 Temperature The method of PID control to achieve the target diameter of the single crystal rod. In these methods, multi-stage PID control for changing the PID constant can be suitably employed. The PID control is one of the feedback control methods, which is a control method in which a signal proportional to the output of the system, an output integrated signal of the system, and an output differential signal of the system are combined. The specific method is to control the pulling (drawing) speed according to the following formula (A), so that the controller preferentially performs the diameter control when the top of the single crystal rod 24 is pulled, and the pulling is preferentially carried out in the later part of the lifting top. speed control:

Vn = Vs+Pkp(Vm/Vt) [ dn+IKi(Vm/Vc)I{(4.,4-dn)/2}x At + DKd(Vm/Vt){(dn—4])/Δί}χ exp(-t/T)〕.........(A) 式中:Vn為應控制之提拉速度;Vn = Vs+Pkp(Vm/Vt) [ dn+IKi(Vm/Vc)I{(4.,4-dn)/2}x At + DKd(Vm/Vt){(dn—4])/Δί }χ exp(-t/T)〕.........(A) where: Vn is the pulling speed that should be controlled;

Vs為設定提拉速度;Vs is the set pulling speed;

Vm/Vt為以矽單結晶之頂部提拉時為基準之PID常 15 1311168 數之梯度; I常數為控制之頂部之P常數,1為PID控制之頂邱 正係數;叫常數常數;以為”數之補 4為此次之目桿正係數w常數之補正係數; 目標直徑與;與貫測直徑之直徑偏差;‘,為前次之 Δ t ^ g* ,杬之直徑偏差;Δΐ為取樣時間;t為〇 IKi(=v間:’ τ為時間常數。又PKP(vm/Vt)為P常數項;' m t為1常數項;DKd(Vm/Vt)為D常數項。、 棒二提拉,結晶 24之直_呈_— 5 G_左右,”單結晶棒 鲁二t楚 左右之範圍A(參與第1圖)。 如下:^ 1及2間說明如此構成之料晶之製造方法 提拉石夕單結晶棒24之頂部時,由於設定提拉逮度快, 直徑變動亦大’故將控制器之(A)式中之p常數值、工常數 值及D常數值設定較大值。即’由於頂部之提拉時常 數之傾度Vxn/Vt大,故P常數項q常數項及〇常數項變 大,結果對直徑偏差之提拉速度之補正量變大且直徑控制 優先,故可使頂部之直徑變動快速及安定。 另外,提拉頂部以後之部份時,由於設定提拉速度慢 直徑變動亦小,故於頂部以後之提拉時,藉由控制器使式(a) 之P常數之梯度變小而使P常數項]常數項數項變 小。由此結果,對直徑偏差之提拉速度之補正量變小,提 拉速度控制乃優先,於是可實行將頂部以後之提拉速度之 16 1311168 變動控制於最小之直徑控制。 另外’在此實施形態中,吉 直仫控制優先之場合,根據 石7早 '乡口日日棒全長设疋之加熱考:w w '现又映射圖像(map)及固液界 =及:液殘篁之各檢測輸出,補正加熱器溫度時,亦 曾e 乍為0又疋棱拉速度之關係式運 :加异。即,於記憶體將根據融液量及固液界面位置之 力口益溫度之補正量作為映射圖像加以儲存,依設定提拉 連度之大小對此補正量進—步補正,或作為設定提拉速度 ::數補正加熱器溫度之邮常數,便可快速的抑制提拉 ^果,即使優先直徑控制,仍可使料結晶棒切割 所传之晶圓全面成為無缺陷狀態。 次說明本發明之第2實施形態於下: 此實施形態係將提拉速度加以pm控制,使石夕單结晶 棒^達到目標直徑之方法及將石夕單結晶棒24之目標直徑 與只測直徑之直猩偏差直接回饋之方法,以及將上述直徑 偏差之變化量作為偏差回.饋至現在之提拉速度之方法予以 、·且&,而將矽單結晶棒24控制成為目標直徑。 日在直徑控制優先時,根據第1實施形態所述之式(A)控 制提拉速度’而提拉速度控制優先時’根據下式⑼控制提 拉速度:Vm/Vt is the gradient of PID often 15 1311168 based on the top of the single crystal of the 矽 single crystal; I constant is the P constant at the top of the control, 1 is the positive coefficient of the PID control; it is called constant constant; The complement of the number 4 is the correction coefficient of the positive coefficient w constant of the target rod; the target diameter and the diameter deviation from the measured diameter; ', is the previous Δ t ^ g*, the diameter deviation of the 杬; Δΐ is the sampling Time; t is 〇IKi (=v: 'τ is a time constant. PKP(vm/Vt) is a P constant term; ' mt is a constant term; DKd(Vm/Vt) is a D constant term.) Lifting, the crystallization of the straight line _ is _-5 G_, "the single crystal rod Lu two t Chu around the range A (involved in Figure 1). The following: ^ 1 and 2 illustrate the manufacture of the crystal material When the top of the crystallized rod 24 is lifted, the setting of the lift is fast and the diameter changes greatly. Therefore, the p constant value, the work constant value and the D constant value in the controller (A) are set. The value is large. That is, because the inclination Vxn/Vt of the top pull constant is large, the q constant term and the 〇 constant term of the P constant term become larger, and the result is a complement to the pull speed of the diameter deviation. The amount becomes larger and the diameter control takes precedence, so that the diameter of the top can be changed quickly and stably. In addition, when the top part of the top is lifted, since the setting of the pulling speed is small, the diameter variation is small, so when lifting at the top, borrowing The gradient of the P constant of the equation (a) is made smaller by the controller, and the term of the constant term of the P constant is reduced. As a result, the correction amount of the pulling speed of the diameter deviation becomes small, and the pulling speed control is given priority. Therefore, it is possible to control the variation of the 16 1311168 speed of the lifting speed at the top to the minimum diameter control. In addition, in this embodiment, the priority of the control of the Zhizhi , is based on the stone 7 early 'township day rod length疋 heating test: ww 'current map image (map) and solid-liquid boundary = and: liquid sputum detection output, when correcting the heater temperature, also e 乍 0 and 疋 拉 pull speed relationship运: Adding the difference. That is, in the memory, the correction amount of the force and the temperature according to the amount of the melt and the position of the solid-liquid interface is stored as a map image, and the correction amount is adjusted according to the magnitude of the set pull-up degree. Correction, or as a setting Degree:: Corrects the mail constant of the heater temperature, and can quickly suppress the pulling effect. Even if the priority diameter is controlled, the wafer transferred by the material crystal rod can be completely turned into a defect-free state. The second embodiment is as follows: This embodiment is a method in which the pulling speed is controlled by pm, so that the stone core single crystal rod reaches the target diameter, and the target diameter of the stone single crystal rod 24 is deviated from the diameter of the measured diameter. The direct feedback method, and the amount of change in the diameter deviation as a method of feeding back to the current pulling speed, and controlling the single crystal rod 24 to the target diameter. When the pulling speed is controlled according to the formula (A) according to the first embodiment and the pulling speed control is prioritized, the pulling speed is controlled according to the following formula (9):

Vr^VrM+P〔dn-UlRi^ + cyqx Μ + 〇 (dn+2dn.r6dn.2+2dn.3 + dn.4)/(6At)] 式中:Vn為此次應控制之提拉速度;Vr^VrM+P[dn-UlRi^ + cyqx Μ + 〇(dn+2dn.r6dn.2+2dn.3 + dn.4)/(6At)] where: Vn is the pulling speed that should be controlled this time ;

Vn-1為前次應控制之提拉速度; 17 Ι3Π168 Ρ為PID控制之Ρ常數;I為PID控制之I常數;D 為PID控制之D常數; dn為此··人之目標直徑與實測直徑之直徑偏差; dn-!為前次之目標直徑與實測直徑之直徑偏差; dn-2為别第二次之目標直徑與實測直徑之直徑偏差; dn·3為4第二次之目標直徑與實測直徑之直徑偏差; dn-4為刖第四次之目標直徑與實測直徑之直徑偏差; △ t為取樣時間。 又,依式(B) ’可將提拉速度控制,使此次應控制之提 拉速度Vn與前次應控制之提拉速度Vn.2之差之絕對值不超 過所定之提拉速度之變動幅度。 爰說明依上述製造矽單結晶之方法於下: 提拉石夕單結晶棒24之頂料,與第1實施形態同樣, 由於设定提拉速度快,直徑變動亦A,因此於控制器設定 气()之P常數、1 *數及D常數為較大。即,因頂部之提 拉日可PID <數之梯度Vm/Vt大,故P常數、I常數及D 常數變大。結果,對直徑偏差之提拉速度之補正量變大, 直徑控制會優先,能使頂部之直徑變動速度安定。 另外,在提拉頂部以後部份時,用式(B)算出此次之提 拉速度便可進-步控财單結晶棒24之提拉速度之變動。 ”體而σ以4 -人之應控制之提拉速度Vn-I為基準之實測 直裎與目私直徑之差之前次的直徑偏差作為占” 1時,以此 次的直徑偏差作為偏差6 η時,以此偏差^減去上述偏差6 Η之值來補正前次之應控制之提拉速度Vy,故能進一步 18 1311168 抑制;ε夕單結晶棒24之提拉速度之變動。 再者於第2貫施形態中,以石夕單結晶棒24之目標直 I /、貝測直控之直控偏差的變化量作為偏差回饋至石夕單結 曰曰棒之提拉速度時,最好是對上述提拉速度實行piD控制, 使不超過對現在之提拉速度之補正之最大變動幅度。具體 而言,f前次之應控制之提拉速度Vni回饋至下一次之應 控=之提拉速度Vn(即此次之應控制之提拉速度)時,最好 對提拉速度設一限制,使提拉速度之變化量不超過最大補 正差Ho,以抑制提拉速度之變動。如此,在進行矽單結晶 棒之頂部以後之提拉時,以前次之⑦單結晶棒之提拉速度 作為基$,將石夕單結晶棒之直捏偏差之變化量作為偏差回 饋至此-人之矽單結晶棒之提拉速度時,若超過對現在之提 拉速度之補正之最大變動幅度之場合,此補正因受最大變 動巾田度之限制’故能將⑦單結晶棒之提拉速度之變動抑 於最小限。 第3〜12圖表示本發明第3實施形態。 此實施形態為’於第“戈2實施形態之矽單結晶棒以 之製造方法中加-依石夕單結晶棒之缺陷模擬法之品質 計算(第3〜5圖及u圓)。即’如第4、5及^圖所示,首 先使用電腦根據總會傳熱解析計算矽單結晶棒2 4之自提拉 開始起每隔一定時間之石夕單結晶棒的溫度分佈。同時,在 石夕單結晶棒之實際提拉中,使用提拉開始起至所定時刻t 止之提拉速度實測特性(profile)及提拉開始至提拉終了止 之設定提拉速度輸人電腦實行妙單結晶棒之品質預測計 19 1311168 算。在預測到矽單結晶棒24有不良部發生時,算出修正不 良。p所用之矽單結晶棒的修正提拉速度及修正加熱器溫 f,而將此等修正提拉速度及修正加熱器溫度回饋至溫度 提拉速度及設定加熱器溫度。 具體而言,當預測到在矽單結晶棒24中會發生侵入型 轉位缺陷時,對設定提拉速度藉增加微小之所定量八5加 以修正,然後根據此修正過之提拉速度,使用電腦依後述 之缺陷模擬方法計算點缺陷之擴散及缺陷之形成。計算後 2測到仍會發生侵入型轉位缺陷之場合,則進一步修正設 定提拉速度使增加,然後根據此修正過之提拉速度, 與上述同樣使用電腦依缺陷模擬方法計算點缺陷之擴散及 缺fe之形成。此計算一直反覆實行至預測到不再發生侵入 型缺陷止。若預測到不再發生侵入型缺陷之場合,以上述 修正過之設定提拉速度替換設定提拉速度,同時將溫度器 設定溫度降低所定之溫度。如此即能減低或阻止矽單結晶 棒之不良部的發生。在此所謂侵入型轉位缺陷(以下稱為 L/DL)係指由過剩之格子間矽凝集形成之轉位缺陷。 再者,若預測到矽單結晶棒24中會發生空隙缺陷(v〇id defects)時則對設定提拉速度藉減少所定量加以修正, 然後根據此修正過之提拉速度,使用電腦依缺陷模擬方法 計算點缺陷之擴散及缺陷之形成。計算後預測到仍會發生 空隙缺陷之場合,則進一步修正設定提拉速度使減少△ 5,然後根據此修正過之提拉速度,與上述同樣使用電腦 依缺陷模擬方法計算點缺陷之擴散及缺陷之形成。此計算 20 1311168 直反覆貝仃至預測到不再發生空隙缺 再發生空隙缺陷之場人,以卜、+w&x 右預測到不 換机過之設定提拉速度替 :楗拉速度,同時將加熱器設定溫度降低所定之溫 此即能減低或阻切單結晶棒之不良部的發生。在 所^空隙缺陷係指由過剩之空孔凝集形成之空隙缺陷。 社曰I早,1晶棒24之品質的預測及修正提拉速度係依石夕單 —棒之缺陷模擬方法計算求出,即用電腦求㈣ =:空隙之濃度分佈及大小分佈之後,再用電腦求::; 早、,Ό曰曰棒内之南溫氧析出物之濃度分佈及大小分佈。 〔I#〕石夕單結晶棒24内之空隙之漠度分佈及大小分佈之 求算法 步驟1 首先,藉由石夕單結晶提拉裝410任意的定義提拉石夕單 結晶棒24時之單結晶製造條件。此單結晶製造條件為使用 提拉速度裝置1G提拉♦單結晶u時,回饋至提拉裝置ι〇 之加熱區(hot zone)之變數依一定間隔予以變量之參數群 Ρΐ Ρΐ.........Ρν。又,此單結晶製造條件包含矽單結晶棒之 提拉速度、⑪單結晶棒之回轉速i,石英㈣15之回轉速 =、氬氣之流量、構成加熱蓋(heatcap)之構件之形狀及材 質、加熱蓋下端與矽融液13表面間之間隙及加熱器之輸出 等。 步驟2 在此步驟疋將石夕單結晶棒24提拉至所定長度L丨(列如 100 nun之狀態時之各構件,即主室(chamber)11、石英坩堝 21 1311168 15、石夕融液13、矽單結晶棒24、黑鉛製内層容器12a、外 層容器(保溫筒)12b等加以網目分割予以模化。具體而言, 將上述加熱區之各個構件之網目點(meshp〇int)之座標資料 輸入電腦中。此時,將矽融液|13之網诏中u之矽單结,晶棒之 徑向之網目且矽融液13之矽單結晶棒正下方之一部分或全 部之網目(以下稱徑向方向網目)設定為〇 〇1〜5〇〇咖,最好 為0.25〜1 _〇〇 min。同時將石夕融液丨3中之網目中之石夕單结晶 棒之長度方向之網目且矽融液13之—部分或全部之網目 (以下稱長度方向網目)設定為〇.〇1〜5.〇〇咖,最好為〇」〜 0.5 mm 〇Vn-1 is the lifting speed that should be controlled before; 17 Ι3Π168 Ρ is the Ρ constant of PID control; I is the I constant of PID control; D is the D constant of PID control; dn is the target diameter and measured by human The diameter deviation of the diameter; dn-! is the diameter deviation between the previous target diameter and the measured diameter; dn-2 is the diameter deviation of the second target diameter and the measured diameter; dn·3 is the second target diameter of 4 Deviation from the diameter of the measured diameter; dn-4 is the diameter deviation of the fourth target diameter from the measured diameter; Δt is the sampling time. Further, according to the formula (B) 'the pulling speed can be controlled so that the absolute value of the difference between the pulling speed Vn to be controlled and the pulling speed Vn. 2 to be controlled before does not exceed the predetermined pulling speed. The extent of the change.爰The method for producing the single crystal according to the above is as follows: The topping material of the single crystal rod 24 of the pulling stone is the same as that of the first embodiment, and since the pulling speed is set to be fast, the diameter is also changed by A, so it is set in the controller. The P constant, the 1* number, and the D constant of gas () are large. That is, since the gradient of the top of the PID can be larger than the gradient Vm/Vt of the number, the P constant, the I constant, and the D constant become large. As a result, the amount of correction for the pulling speed of the diameter deviation becomes large, and the diameter control is prioritized, so that the diameter of the top can be stabilized. In addition, when the top portion of the top of the lift is lifted, the pulling speed of the crystal rod 24 can be controlled by the formula (B) to calculate the pulling speed of the bill. "The body and σ are measured by the pull-up speed Vn-I of 4 - person, and the difference between the measured diameter and the private diameter of the previous diameter is taken as "1", and the deviation of the diameter is used as the deviation 6 In the case of η, the value of the deviation of 6 Η is subtracted from this deviation to correct the previous pulling speed Vy to be controlled, so that it can be further suppressed by 18 1311168; the fluctuation of the pulling speed of the single crystal rod 24 is obtained. Further, in the second embodiment, the amount of change in the direct control deviation of the direct I / / direct measurement of the Shi Xi single crystal rod 24 is used as the deviation to the pull speed of the Shi Xi single knot rod. Preferably, the piD control is performed on the above-mentioned pulling speed so as not to exceed the maximum variation of the correction of the current pulling speed. Specifically, when the pulling speed Vni of the previous control should be fed back to the next pulling speed Vn (ie, the pulling speed to be controlled), it is preferable to set the pulling speed. The limit is such that the amount of change in the pulling speed does not exceed the maximum correction difference Ho to suppress the variation of the pulling speed. In this way, when lifting the top of the single crystal rod, the pulling speed of the previous 7 single crystal rod is used as the base $, and the amount of change in the straight pinch deviation of the stone single crystal rod is fed back to the person as a deviation. In the case of the pulling speed of the single crystal rod, if the maximum fluctuation range of the current lifting speed correction is exceeded, the correction can be lifted by the 7 single crystal rod due to the limitation of the maximum variation. The change in speed is limited to the minimum. Figures 3 to 12 show a third embodiment of the present invention. This embodiment is a quality calculation of the defect simulation method of the 矽 夕 单 single crystal rod in the manufacturing method of the “ 结晶 2 实施 实施 实施 第 第 第 第 第 第 第 第 第 第 第 第 第 第 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 As shown in Figures 4, 5 and ^, firstly, using the computer to calculate the temperature distribution of the single crystal rod of the single crystal of the single crystal rod 24 from the start of the lifting according to the total heat transfer analysis, at the same time, In the actual lifting of the Shixi single crystal rod, the measured profile of the pulling speed from the start of the lifting to the predetermined time t and the setting of the lifting speed to the end of the lifting of the pulling speed are input to the computer. The quality of the crystal rod is predicted to be 13 1311168. When it is predicted that the defective single crystal rod 24 has a defective portion, the correction failure is calculated. The corrected pulling speed of the single crystal rod used for p and the corrected heater temperature f are used. Then, the lifting speed is corrected and the heater temperature is fed back to the temperature pulling speed and the heater temperature is set. Specifically, when the intrusion type indexing defect is predicted to occur in the single crystal rod 24, the pulling speed is set. By adding a small amount of quantification 5 to correct, and then according to the modified pull speed, use the computer to calculate the diffusion of defects and the formation of defects according to the defect simulation method described later. After the calculation 2, the intrusion type transposition defect still occurs, then further Correctly set the lifting speed to increase, and then according to the modified pulling speed, use the computer to calculate the spread of the defect and the lack of the defect according to the defect simulation method. This calculation is repeated until the prediction is no longer intrusion. If the intrusion type defect is predicted to occur, replace the set pulling speed with the above-mentioned corrected pulling speed, and lower the temperature set by the temperature. This will reduce or prevent the order. The occurrence of a defective portion of the crystal rod. Here, an invasive translocation defect (hereinafter referred to as L/DL) refers to an index defect formed by agglomeration of excess inter-column. Further, if a single crystal rod 24 is predicted In the case of void defects (v〇id defects), the set pull speed is corrected by the amount of reduction, and then the corrected pull speed is corrected. Degree, using the computer to calculate the diffusion of the defect and the formation of the defect according to the defect simulation method. When the calculation predicts that the void defect still occurs, the lifting speed is further corrected to reduce Δ 5, and then the modified lifting is performed according to the correction. Speed, as in the above, use the computer to calculate the diffusion of defects and the formation of defects by the defect simulation method. This calculation 20 1311168 directly reverses the Bellow to the person who predicts that there will be no voids and then the void defect occurs, to the +, w&;x Right predicts that the lift speed will not be changed for the machine: the pull speed, and the temperature set by the heater can be lowered to reduce or hinder the occurrence of the bad part of the single crystal rod. Defect refers to the void defect formed by the agglomeration of excess pores. The prediction of the quality of the 1st rod 24 and the correction of the pulling speed are calculated according to the simulation method of the stone-single-rod defect, that is, the computer Seek (4) =: After the concentration distribution and size distribution of the voids, use the computer to find::; Early, the concentration distribution and size distribution of the oxygen precipitates in the south of the crowbar. [I#] The distribution of the indifference and the size distribution of the voids in the Shixi single crystal rod 24 is as follows: First, by using the Shixi single crystal pulling device 410, the definition of the Tilatite single crystal rod 24 is arbitrary. Single crystal manufacturing conditions. The single crystal manufacturing condition is that when the pulling speed device 1G is used to lift the single crystal u, the variable of the hot zone fed back to the pulling device ι is given a parameter group of variables at a certain interval. .....Ρν. Further, the single crystal production conditions include the pulling speed of the single crystal rod, the rotational speed i of the single crystal rod, the rotational speed of the quartz (four) 15, the flow rate of the argon gas, and the shape and material of the member constituting the heat cap. The gap between the lower end of the lid and the surface of the crucible 13 and the output of the heater. Step 2 In this step, the stone singular crystal rod 24 is pulled to a predetermined length L 丨 (the column is in the state of 100 nun, that is, the main chamber (chamber) 11, quartz 坩埚 21 1311168 15 , Shi Xirong 13. The single crystal rod 24, the black lead inner container 12a, the outer container (insulation tube) 12b, etc. are meshed and molded. Specifically, the mesh points of the respective members of the heating zone are (meshp〇int) The coordinate data is input into the computer. At this time, the 矽 矽 | 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 (hereinafter referred to as radial direction mesh) is set to 〇〇1~5〇〇, preferably 0.25~1 _〇〇min. At the same time, the length of Shixi single crystal rod in the mesh of Shi Xi Rong 丨3 The mesh of the direction and the mesh of some or all of the melt 13 (hereinafter referred to as the length direction mesh) are set to 〇.〇1~5.〇〇, preferably 〇"~ 0.5 mm 〇

將徑向之網目設定〇.〇丨〜孓⑼咖範圍之理由為,若O N 伽1以下時,計算所需時間極長,超過5 〇〇咖時,計算變為 不安定,即使反覆計算,固液界面形狀仍難定於一定。又, 將長度方向之網目限定於Ο.ίΗ咖以下時,計算所需時間極 長超過5.00圆1時,固液界面形狀之計算值會與實測值不 :致。另外,將徑方向網目的一部份限定於〇〇1〜5〇〇卹 範,之場合,最好料單結晶棒正下方之料液13中靠= 矽單結晶棒之外周緣部之矽融液13限定於上述範目,同= =長度方向之網目的一部份限定於〇·〇1〜5 〇〇咖範圍^尸 s,最好將矽融液之液面附近及底部附近限定於上述範野 步驟3 m ° k在此按上述加熱區之各構件整理其網目並對此整理 ’’’罔目將各構件之物性值分別輸入電腦中。例如,若主^ 是不銹鋼製時,則將供不銹鋼之熱傳導率、幅2、至11 田柯平、枯性 22 1311168 率、體積膨脹係數、密度及比熱輸人電腦中。另外,將石夕 早結晶棒24之提拉速度及對應於此提拉速度之⑦單結晶棒 Μ之提拉速度,及後述之亂流模式⑺之礼流參數 腦中。 电 步驟4 …=此使用電腦,根據加熱器之發熱量及各構件之幅射 率計算求出加熱區之各構件之表面溫度分佈。即,任意設 定加熱器之發熱量,將其輸入電腦之同時,利用電腦,從 各構件之幅射率計算各構件之表面溫度分佈。 步驟5 在此根據加熱器之各構件之表面溫度分佈及熱傳導 率。使用電腦解熱傳導方程式(1)求出各構件之内部溫度分 佈。在此,為了方便說明,使用xyz直交座標系,實際計 算則使用圓筒座標系: A d T d β Q -- .aT Αχ Λ + 9 ί, θτ + 9 ίλ 3T) at a t d y 卜a y J d Z 卜az J + q (1) 式中;p為各構件之密度,C為各構件之比熱,Τ為各 構件之各網目點之絕對溫度’ t為時間、;lx、又7及。為 各構件之熱傳導率之X,y & ζ方向成分,q為加熱器之發 熱量。 至於,石夕融液13則先由上述之熱傳導方程式(丨)求出矽 融液13之内部溫度分佈後’根據此矽融液13之内部分佈 連結假定矽融液13為亂流所得之亂流模式(2)及粘性流體 23 1311168 方程式(Navie-Stokes equation)(3)〜(5) ’利用電腦求算矽 融液13之内部流速分佈·· K t~ —— X X C X d Vic ...... ( 2 ) P r t 式中;kt為矽融液13之亂流熱傳導率,c為矽融液之 比熱,Prt為普朗特數(prandti number),p為石夕融液13之密 度,c為亂流參數’ d為距離儲存矽融液13之石英掛禍 壁之距離,k為對矽融液13之平均流速之變動成分之平方 和: d u ( d u^7+u — v,+w.3u d x d y d z —+ ( VI + Vt) p dxThe reason why the radial mesh is set to 〇.〇丨~孓(9) coffee range is that if ON gamma 1 or less, the calculation time is extremely long. When it exceeds 5 〇〇 coffee, the calculation becomes unstable, even if it is calculated repeatedly. The shape of the solid-liquid interface is still difficult to determine. In addition, when the mesh in the length direction is limited to Ο. Η Η Η , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , In addition, when a part of the radial direction mesh is limited to the 〇〇1~5 范 范 , , , , , , 〜 〜 〜 〜 〜 〜 〜 料 正 正 正 正 正 正 正 正 正 正 正 正 正 13 13 13 The melt 13 is limited to the above-mentioned specifications, and the part of the net with the == length direction is limited to 〇·〇1~5 〇〇 范围 ^ 尸 , , , , , , , , , , , 限定 限定 限定 限定 限定In the above-mentioned Fanye step 3 m ° k, the meshes of the components in the above-mentioned heating zone are arranged and the physical property values of the components are respectively input into the computer. For example, if the main ^ is made of stainless steel, the thermal conductivity of the stainless steel, the width 2, the 11 to the Kopping, the dryness 22 1311168 rate, the volume expansion coefficient, the density and the specific heat are input into the computer. Further, the pulling speed of the early crystallization rod 24 and the pulling speed of the 7 single crystal rod corresponding to the pulling speed, and the ritual flow parameter of the turbulent flow pattern (7) to be described later. Electric Step 4 ...= This uses a computer to calculate the surface temperature distribution of each component of the heating zone based on the heat generated by the heater and the radiation rate of each component. That is, the heat generation amount of the heater is arbitrarily set, and when it is input to a computer, the surface temperature distribution of each member is calculated from the radiation rate of each member by a computer. Step 5 Here, according to the surface temperature distribution and thermal conductivity of each member of the heater. The internal temperature distribution of each member was obtained using the computer's heat dissipation equation (1). Here, for convenience of explanation, the xyz orthogonal coordinate system is used, and the actual calculation uses the cylindrical coordinate system: A d T d β Q -- .aT Αχ Λ + 9 ί, θτ + 9 ίλ 3T) at atdy ay y J d Z 卜 aj J + q (1) where p is the density of each component, C is the specific heat of each component, and 绝对 is the absolute temperature of each mesh point of each component 't time; lx, 7 and. For the thermal conductivity of each component, X, y & ζ direction component, q is the heat generated by the heater. As for the Shi Xi Rong liquid 13, the heat transfer equation (丨) is used to determine the internal temperature distribution of the sputum melt 13 'according to the internal distribution of the mash liquid 13 and the turbulent flow 13 is assumed to be turbulent flow. Flow mode (2) and viscous fluid 23 1311168 Equation (Navie-Stokes equation) (3) ~ (5) 'Using the computer to calculate the internal flow velocity distribution of the sputum 13 · K t~ —— XXCX d Vic ... ( 2 ) P rt where kt is the turbulent heat transfer rate of sputum melt 13 , c is the specific heat of sputum melt, Prt is the prandti number, and p is the stone smelt 13 Density, c is the turbulent flow parameter 'd is the distance from the quartz wall of the sputum melt 13 and k is the sum of the squares of the variation of the average flow velocity of the mash 13: du ( du^7+u — v ,+w.3u dxdydz —+ ( VI + Vt) p dx

a2u a2u a2u ,Fx d X2 dy2 a z2 J 十 P (3) d v . d v u dx ''dy 1 dp v^L+wav d z (V 1 + V t )A2u a2u a2u , Fx d X2 dy2 a z2 J 十 P (3) d v . d v u dx ''dy 1 dp v^L+wav d z (V 1 + V t )

d-~w ~dT -u p d y 3 w . ... 3 w , 3 w ——+ V-—+w—— d x d y d z -----h( vi +vt) p d z a2v a2v 52 v ' .Fy ,.、 lax2 1 3 y2 5 Z2 j 丁 -- P . ”,·…D-~w ~dT -updy 3 w . ... 3 w , 3 w ——+ V--+w——dxdydz -----h( vi +vt) pdz a2v a2v 52 v ' .Fy , ., lax2 1 3 y2 5 Z2 j Ding -- P . ", ·...

d 2w d 2w θ 2w 1 ;F z .3 x 2 d y2 a z2 ; I (5) 式中:11、v及w為矽融液13之各網目點之流速之x、 y及z方向成分,Vl為矽融液之分子動粘性係數(物性值), vt為因矽融液之亂流效果之動粘性係數,Fx、Fy及Fz為作 用於石夕融液之體積力(bady force)之x、y及z方向成分。 24 1311168 上述m式(2)稱為KL_模 最好是用0.4〜0.6範圍肉々/立± 衩式之亂流參數c „ ^ Λ _ 圍内之任思值。將亂流參數C限定名 〇.4〜〇·ό範圍是,因疋為 界面报壯合古6時,由計算所得之 界面形狀會有不與實測值一 侍之 性流體方程式(3)〜(5)係把—々Λ 又,上述之特 I )係叙疋矽融液為非壓縮性且 一定之流體時之運動方程式。 X為 根據上述求得之石夕融液13之内部流速分佈解 式6,即可進-步利用電腦求出料融液之對流列:考^ 矽融液之内部溫度分佈: (6) 3Ύ , 5Τ 5Τ U-T--Η V -Γ-+ W-— d x d y d z 'd2τ a2τ d2T ,~dxY+~dVT+~d~z^ 式中:U、w為矽融液13之各網目點之流速之x、 yAz方向成分’ T為㈣液之各網目點之絕對溫度,々為 石夕融液之密度,融液之比熱,分子熱傳導率(物 性值)’ kt為用式(1)計算之亂流熱傳導率。 步驟6 在此,用電腦根據第11圖之含有點3所示之矽之三交 會點〔固體、液體及氣體之三交會點(tri_juncti〇n)〕之等溫 值計算矽單結晶棒24及矽融液13之固液界面形狀。 步驟7 在此’改變(逐漸增大)輸入電腦之加熱器的發熱量,同 時反覆上述第4步称至第6步驟,直至三交會點$ .達到石夕 單結晶棒24之融液之後,計算提拉裝置1〇内之溫度分佈, 25 1311168 =β夕早—之網目之座標及溫度而將該等資料記憶於電 步驟8 在此’於石夕單結晶棒24之提拉長度^加上占(例如⑽ 麵)同時反覆上述步驟2至步驟7之操作後,計算提拉裝置 10内之溫度分佈求㈣單結晶棒之網目之座標及溫度^將 該等資料記憶於電腦中。此步驟8 一直進行至石夕單結晶棒 冷部完了止,即矽單結晶棒之提拉長度L1達到U(L2為脫 離夕融液13時之石夕單結晶棒之長度,即成長完了時之結晶 長度)而矽單結晶棒脫離矽融液之後,進一步提拉矽單結晶 棒,使其鬲度Ηι(Ηι為矽單結晶棒之直胴開始部至矽融液 之液面之距離(參考第6圖)達到H2(H2為冷卻完了時之直 胴開始部㈣融液之液面之聽)止。另外,在料結晶棒 脫離石夕融夜後’求石夕單結晶棒之提拉高度m加上5 (例如 50醜),而與上述同樣反覆步驟2至步驟了之操作。 當矽單結晶棒24之提拉速度Hl達到Η?時,即進行步 驟9。 步驟9 在此’將石夕融液中成長之矽單結晶棒從融液開始提拉 之時間to至;ε夕單結晶棒脫離石夕融液後,再提拉至冷卻完了 時之時間ti ’依所定之時間間隔△(秒(微小時間間隔)加以 區割。此時不但將矽單結晶棒内之格子間矽及空孔之擴散 係數及境界條件,同時將後述之求算 空隙(void)及高溫氧析 出物之濃度分佈及大小(以找)分佈之式所用之定數分別輸入 26 1311168 電月旬中爾後私上述區割之時間間隔△ ί秒,由步驟8求得 之矽單曰結晶棒之網目之座標及溫度之資料計算矽單結晶^d 2w d 2w θ 2w 1 ; F z .3 x 2 d y2 a z2 ; I (5) where: 11, v and w are the x, y and z-direction components of the flow velocity of each mesh point of the mash 13 Vl is the molecular dynamic viscosity coefficient (physical property value) of the sputum melt, vt is the dynamic viscosity coefficient of the turbulent flow effect of the sputum melt, and Fx, Fy and Fz are the bady force acting on the Shixi melt. The x, y and z direction components. 24 1311168 The above m formula (2) is called KL_mode. It is best to use the 0.4~0.6 range of meat 々/立± 之 turbulent flow parameter c „ ^ Λ _ within the circumference of the value. The turbulent flow parameter C qualified name The range of 〇.4~〇·ό is that, because the interface is reported to be the same as the ancient 6th, the calculated interface shape will have a fluid equation that does not match the measured value. (3)~(5) Λ Further, the above special I) is a equation of motion when the fluid is non-compressive and a certain fluid. X is the internal flow velocity distribution of the Shixi melt 13 obtained according to the above, and can be advanced. - Step to use the computer to find the convection column of the melt: test the internal temperature distribution of the melt: (6) 3Ύ, 5Τ 5Τ UT--Η V -Γ-+ W-- dxdydz 'd2τ a2τ d2T , ~dxY+ ~dVT+~d~z^ where: U, w is the flow rate of each mesh point of the smelting liquid 13 and the yAz direction component 'T is the absolute temperature of each mesh point of the liquid (4), and the 々 is the stone immersion liquid Density, specific heat of melt, molecular thermal conductivity (physical value)' kt is the turbulent thermal conductivity calculated by equation (1). Step 6 Here, use computer according to Figure 11 containing point 3 The isothermal value of the three intersection points (tri_juncti〇n of solid, liquid and gas) is calculated as the solid-liquid interface shape of the single crystal rod 24 and the crucible melt. Step 7 Here, 'change (increasing Enter the heat of the heater of the computer, and repeat the above steps from step 4 to step 6, until the three intersections point $. After reaching the melt of the stone singular crystal rod 24, calculate the temperature distribution in the 〇1 of the pulling device. , 25 1311168 = β 夕 early - the coordinates of the mesh and the temperature and the data are stored in the electrical step 8 where the 'lifting length of the stone singular single crystal rod 24 plus plus (for example, (10) surface) After the operation of steps 2 to 7, the temperature distribution in the pulling device 10 is calculated. (4) The coordinates and temperature of the mesh of the single crystal rod are memorized in the computer. This step 8 is carried out until the single crystal rod of Shixi When the cold part is finished, the pulling length L1 of the single crystal rod reaches U (L2 is the length of the single crystal rod of the Shixi single crystal when it is separated from the night melt, that is, the crystal length when the growth is completed) and the single crystal rod is separated from the crucible. After the melt is melted, the single crystal rod is further pulled, The degree of Ηι (Ηι is the distance from the beginning of the straight 胴 of the single crystal rod to the liquid surface of the 矽 ( (refer to Fig. 6) reaches H2 (H2 is the beginning of the straight enthalpy when the cooling is completed (4) In addition, after the material crystal rod is separated from Shi Xirong night, the lifting height m of the stone single crystal rod is added by 5 (for example, 50 ugly), and the operation of step 2 to step is repeated as described above. When the pulling speed H1 of the single crystal rod 24 reaches Η?, the step 9 is performed. Step 9 Here, the time from the start of pulling the single crystal rod grown in the Shixi melt is started; After the ε 夕 single crystal rod is separated from the Shi Xirong liquid, it is pulled up to the time when the cooling is completed, and the time interval Δ is determined according to the determined time interval Δ (second (minor time interval)). At this time, not only the diffusion coefficient and the boundary condition of the lattice and the pores in the single crystal rod, but also the distribution of the void and the concentration distribution (seek) of the high-temperature oxygen precipitate described later will be described. The number used is input to 26 1311168, and the time interval between the above-mentioned zones is Δ ί sec. The data of the coordinates and temperature of the mesh of the single-crystal crystallization rod obtained in step 8 is calculated as 矽 single crystal ^

24之提拉長度Ll及提拉高度Ηι以及石夕單結晶棒内之温产 分佈。 X 即,在步驟2〜8,依每一提拉長5求出矽單結晶之網 目之座標及溫度。由於例如提拉石夕單結晶%腿需時數十分 鐘,則將此數十分鐘之矽單結晶之網目之溫度變化作為時 間之函數加以微分’求出t(^At秒後之碎單結晶棒之提拉 長度U及提拉高度Hl以及矽單結晶棒内之溫度分佈。 步驟10 繼之,根據矽單結晶棒内之空孔及格子間矽之擴散係 數及境界條件解擴散方程式(後4之式(9)、(10)及(13)),即 可求出經過At秒後之空孔及格子間隙之濃度分佈。 具體έ之’空孔之濃度Cv之計算用下式⑺,格子間石夕 之濃度Ci之計算用式⑻。在式⑺、⑻中,為了計算濃度 Cv及Ci之經時(隨時間)的變化,將空孔及格子間矽之熱$ 衡假設為在矽單結晶之全表面維持。 (8) C i e = K 2 exp__马 i . kTj 式中,Κι及K2為常數(c〇nstant),艮及Ev分別代表格 子間石夕及空孔之形成能量,Cve&〜之e代表平衡量,κ 代表波子曼常數’ Τ為絕對溫度。 27 1311168 上述平衡式以時間微分,則對空孔及格子間矽分別成 下式(9)及式(1〇)。 d C ’ 'The pull-up length Ll of 24 and the height of the pull-up Ηι and the temperature distribution in the Shixi single crystal rod. X, that is, in steps 2 to 8, the coordinates and temperature of the net of the single crystal are obtained for each lift length 5. Since, for example, the tiling single crystal % leg takes several tens of minutes, the temperature change of the mesh of the tens of minutes of the single crystal is differentiated as a function of time to find the singular crystal of t(^At seconds) The pull length U and the lift height H1 of the rod and the temperature distribution in the single crystal rod. Step 10 Next, according to the diffusion coefficient and the boundary condition of the pores in the single crystal rod, the diffusion equation and the boundary condition (after 4 (9), (10), and (13)), the concentration distribution of the pores and the lattice gap after At seconds is obtained. The specific calculation of the concentration Cv of the pores is as follows (7). In the equations (7) and (8), in order to calculate the change of the concentration Cv and Ci over time (with time), the heat balance between the pores and the lattice is assumed to be in the equation (7) and (8). (8) C ie = K 2 exp__马 i . kTj where Κι and K2 are constants (c〇nstant), and 艮 and Ev represent the formation of the space between the lattice and the void Energy, Cve &~e represents the balance, κ represents the waveman constant ' Τ is the absolute temperature. 27 1311168 Balanced differential time, the pores and the inter-lattice silicon to each. D C of the formula (9) and (1〇) ''

~~^ — — V(DvV Cv)+V — D v C v V T~~^ — — V(DvV Cv)+V — D v C v V T

dr l kBT2 J + 0(T-Tn)kiv(T)(CiCv-Cie(T)Cve(T)) —Θ(Τν〇 - T )Nv〇x 47ΓΓ v〇Dv(Cv—Cve) -©(TP-T)NPx47rRp*/D〇D〇 ......(9 )Dr l kBT2 J + 0(T-Tn)kiv(T)(CiCv-Cie(T)Cve(T)) —Θ(Τν〇- T )Nv〇x 47ΓΓ v〇Dv(Cv—Cve) -©( TP-T)NPx47rRp*/D〇D〇...(9)

~~=-V(DiVCi)+v[-^-DiCiVT (kBT2 + 0(T-Tn)kiv(T)(CiCv-Cie(T)Cve(T)) -0(Tv〇-T)Nv〇x47rrv〇Di(Ci-Cie) ......(10) 式中· Θ(Χ)為赫維赛德函數(Heaviside function)。即, x<0時Θ〇〇=〇,且x>〇時Θ(Χ)=Μβ几代表高溫氧析出 物之形成開始溫度τρ與空孔之形成開始溫度TvG比較時較 N 一方之溫度。又,式(9)及(10)之各右邊之第1項為菲克 擴散式(Fick equation),右邊第1項中之队及仏係由下式 (11)及(12)代表之擴散係數 D v = d ve.xp 'ΔΕν D i = d exp~~=-V(DiVCi)+v[-^-DiCiVT (kBT2 + 0(T-Tn)kiv(T)(CiCv-Cie(T)Cve(T)) -0(Tv〇-T)Nv〇 x47rrv〇Di(Ci-Cie) (10) where Θ(Χ) is the Heaviside function. That is, x<0 Θ〇〇=〇, and x> 〇时Θ(Χ)=Μβ represents the temperature of the N-side when the formation temperature τρ of the high-temperature oxygen precipitate is compared with the formation start temperature TvG of the void. Further, the right side of each of the equations (9) and (10) One item is the Fick equation, and the team and the lanthanide in the first item on the right are represented by the following equations (11) and (12). The diffusion coefficient D v = d ve.xp 'ΔΕν D i = d exp

f ΔΕ j "kT (11) (12) 式中:ΔΕν及AEi分別代表空孔及格子間矽之活性化能 罝,dv及山分別代表常數。式(9)及(1〇)之各右邊的第2項 28 1311168 中之EV及Eit分別代表因熱擴散 之活性化能量;kB為波子曼常數。又,=孔及格子間石, —:數it,為空孔及格子間石夕對㈣—)之再 及⑽之各右邊的第4項中之Nv。為空孔 < /晨度’ Γν。為空孔之半徑,又洼 為高溫氧析出物之濃产 邊的第5項之乂 ς·η 4 /辰度Rp為兩溫氧析出物之半徑,γ為 費量析出物能不變歪的析出所需要之每-氧原子之空孔消 為析^空ΐ式ΐ可成立之ί件為:Dv(C「C/)㈣C。’即 位質量:體::遠足夠的高至能填補Si基質及Si〇2之每單 、體積差之場合。上述以外之場合,下式⑽可成立: ^^-▽(DvVCvHvf^^vCvVTj + ©(T-Tn)kiv(T)(CiCv-Ci-(T)Cve(T)) ~0(Tv〇-T)Nv〇X47rrv〇Dv(Cv-Cve) -NPx 4ttRpDv( Cv-Cve) ……(13) 少哪11 、在此,根據解上述擴散方程式(13)所得之空孔之讀 Cv刀佈,由下式(14)求出空孔之形成開始溫度Tv0 : Τ’ k C. C, -kB°·f ΔΕ j "kT (11) (12) where: ΔΕν and AEi represent the activation energy of the pores and the lattices, respectively, and dv and mountains represent constants, respectively. The second item on the right side of equations (9) and (1) 28 EV and Eit in 13 1311168 represent the activation energy due to thermal diffusion, respectively; kB is the waveman constant. Also, the = hole and the inter-grid stone, -: the number it is the empty hole and the Nv in the fourth item on the right side of the (10) and (10). Empty hole < / morning degree Γ ν. The radius of the void is the fifth term of the rich side of the high temperature oxygen precipitate. η 4 / Chen Rp is the radius of the two-temperature oxygen precipitate, and γ is the amount of the precipitate. The pores of each oxygen atom required for the precipitation are eliminated. The 件 ΐ ΐ ΐ D D D D D D D D D D D D D D D D ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' For the case where the Si matrix and Si〇2 are different in volume and volume, the following formula (10) can be established: ^^-▽(DvVCvHvf^^vCvVTj + ©(T-Tn)kiv(T)(CiCv-Ci -(T)Cve(T)) ~0(Tv〇-T)Nv〇X47rrv〇Dv(Cv-Cve) -NPx 4ttRpDv( Cv-Cve) (13) Less than 11, here, according to the solution The cavity Cv knife cloth obtained by the diffusion equation (13) is obtained by the following formula (14) to determine the formation start temperature Tv0 of the hole: Τ' k C. C, -kB°·

E T,E T,

0 . 6 8 cr v 1 · 5 P + Tv〇0· skB0· 5 Ev ……(14) 式中 cvm為矽融液在融點Tm時之空孔平衡濃度,Ev 29 1311168 為空孔形成能量,Tm為矽單結晶棒之融點溫度。又,〜為 之界面能量’ p騎單結晶棒之 密度,κΒ為波子曼常數。 步驟12容後再說明,在此先說明步驟Β 步驟13 網目的格子點之溫度逐 Τν0時’依下示之近似式 在此’當矽單結晶棒内之各個 漸降低而達到空孔之形成開始溫度 (15)求算空孔之濃度nv0 : N v〇 = a dT ' d t , (DvkBT2) -1· 5(CV) -0. 5 (15) 步驟14 在此’用下綱求出石夕單結晶棒内之各個 點之溫度較空孔之形成開始溫度Tvq為低時之空孔的半徑 rv0 *0 . 6 8 cr v 1 · 5 P + Tv〇0· skB0· 5 Ev (14) where cvm is the equilibrium concentration of the pores at the melting point Tm, and Ev 29 1311168 is the pore forming energy. , Tm is the melting point temperature of the single crystal rod. Further, ~ is the interface energy 'p rides the density of the single crystal rod, and κ Β is the waveman constant. Step 12 and then explain, first explain the steps Β Step 13 When the temperature of the grid point of the grid is Τν0', the following approximation is shown here: when the 矽 single crystal rod is gradually reduced, the pore formation is achieved. Starting temperature (15) Find the concentration of the hole nv0 : N v〇 = a dT ' dt , (DvkBT2) -1· 5(CV) -0. 5 (15) Step 14 Here, use the following outline to find Shi Xi The temperature of each point in the single crystal rod is smaller than the radius rv0 of the hole when the formation start temperature Tvq is low.

P {Dv(Cv — Cve)—Di(Ci - Cie)}dt + .(1 6) 式中:t!代表矽單結晶棒之網目的格 空孔之形成開始溫度Tv。時之時間,r二,,…皿度降低至 繼之反覆實行上述步驟9〜11、13及':工孔之臨界直徑。 土, 及14 ’直至石夕單έ士 a接 達到綱〜1 〇帆之間之特定值,例如冷卻至 驟17)。 c从下(步 上式(9)〜式(16)係藉由電腦計瞀 ^衣出。至於步驟15及 30 ulll68 16則容後說明。 教度佈〕矽單結晶棒24内之高溫氣析出物之濃度分佈及 步驟12 戶fi· 此回到步驟12 ’根據解上述擴散方程式⑼及(1〇) ^付之空孔之濃度Cv分佈,依下式⑽計算求出高溫氧 出物之形成開始溫度TP :P {Dv(Cv - Cve) - Di(Ci - Cie)}dt + . (1 6) where: t! represents the formation start temperature Tv of the void of the mesh of the single crystal rod. At the time of the day, r 2, ..., the degree of the dish is reduced to the above steps 9 to 11, 13 and ': the critical diameter of the hole. Soil, and 14 ' until the stone singular gentleman a connect to reach a specific value between the ~ 1 〇 sail, for example, cool to step 17). c from the bottom (steps (9) ~ (16) by computer 瞀 ^ clothing. As for steps 15 and 30 ulll68 16 will be explained later. Teaching cloth] 矽 single crystal rod 24 in the high temperature gas The concentration distribution of the precipitates and the step 12 are fi; returning to step 12', according to the concentration Cv distribution of the pores obtained by solving the above diffusion equations (9) and (1〇), the high temperature oxygen product is calculated according to the following formula (10). Forming the starting temperature TP:

Tpl· 5 k b 1 * 5 l〇g^ [Co ) Cv y' _ ks0* 5(E〇 + y Ev) L .Com, .C vm , j Tm J + Τ kB〇· 5(E〇 + 7Ev)Tpl· 5 kb 1 * 5 l〇g^ [Co ) Cv y' _ ks0* 5(E〇+ y Ev) L .Com, .C vm , j Tm J + Τ kB〇· 5(E〇+ 7Ev )

•(IV _ Ο . 6 σ, Ρ 式中· C〇代表氧濃度,為細矽融液η之融點h 礼平衡濃度,eg魏溶解能量,Ev為空孔形成能量, 2早結晶棒内之⑴與叫之界面能量,而丫為Si02析 :不變正的析出所需要之每一氧原子之空孔消費量,其 值為0.68。 步驟15 在此,當矽單杜B接〜 、° aa棒内之各個網目的格子點之溫度逐 (ι::而達到间’皿氧析出物之形成開始溫度丁,時,依下式 _求出高溫氧析出物之濃度Np,式中&為常數: ................ —........ ................ ......... NP== a 2 d 丁、1 g —------------------------------------------------------- d t• (IV _ Ο . 6 σ, Ρ where · C 〇 represents the oxygen concentration, is the melting point of the fine 矽 melt η h equilibrium concentration, eg Wei dissolved energy, Ev is the pore forming energy, 2 early crystal rod The (1) and the called interface energy, and the 丫 is the SiO 2 analysis: the hole consumption of each oxygen atom required for the precipitation of invariant positive value, the value is 0.68. Step 15 Here, when the single Du B is connected to ~, ° The temperature of the grid points of the respective nets in the aa bar is (i:: and the temperature at which the formation of the oxygen precipitates in the chamber is reached, and the concentration Np of the high-temperature oxygen precipitates is determined according to the following formula, where & Is a constant: ................ —................................ ... NP== a 2 d D, 1 g —------------------------------------ ------------------- dt

Cv(D oCqIcbT2)-1. 5 ...... (18) 31 1311168 步驟16 「Μ 1 y)衣出石夕單結晶榉内之各個網目的格早 ^度較高溫氧析出物之形成開始溫度= 氧析出物之半徑Rp : 々低矸之冋溫 R, ~Co J Dod t +R, μ t2 19) ,又,上述式(19)可成立之條件為Dv(Cv—Cv& rD〇c :為析出空孔之流速足夠的高至能填補si&fU办之每C 早位質量之體積差之場合,下式(2〇)可成立: RP = if D v( C V Cve)dt+Rcr2 ……(2 0) [r β Λ J 步專 % 17 在此,反覆上述步驟9、10、12、15及16直至矽單結 晶棒冷卻至800〜100(rc之間之特定值,例如9〇(rc以下。 上述式(9)〜(13)及(1 7)〜(20)係藉由電觸計算求出。 步驟18 在此,將上述南溫氧析出物存在之P帶與空孔及高溫 氧析出物皆無存在之無缺陷領域加以劃分,並依計算求出 第1等濃度線HC〗X ’以劃分無缺陷領域比p帶更靠近單結 32 1311168 晶底部側之位置(參照第9圖)。 步驟19 在此,劃分高濃度氧析出物存在之 同時依計算求出坌,八&& d 9久,,,、缺陷領域 ί 料吨,以劃分無缺陷領域比£ 罪、早、〜S日頂部側之位置。上述第丨分佈線係扑 :矽:度對矽融點時之格子間矽之平衡濃度在m〜二3 特定值,例如〇.126%之等濃度線。在此所謂格 子間h農度(M2〜〇.13%之範圍之特定值,例如〇126%, 係指經由實行模擬所得之格子时漠度之巾,從使用壽命 (hfetlme)之標圖(maP)等觀察之對應於B帶之濃度。又 謂B帶係指格子間矽之凝聚體成為核而通過熱處理引起氧 析出物變成高濃度之領域。 步驟20 在此,求出第1等濃度線HC1X之受面點之最大值與第 1分佈線BClxi變曲點之最小值之差若是第i等 濃度線HC1X之變曲點例如有三個時,將其各變曲點 Ql(l,x)、Ql(2,x)及 Ql(3,x)之座標設為〔r Q1(1,x),z Qi(l’x)〕、〔 r Q1(2,x) ’ z Q1(2,x)〕及〔r Q1(3,x),z Q1(3,x)〕, 並求出此等座標中之z座標,即z Q1(i,x)、z Q1(2,x)& z qi(3,x) 中之最大值ZQi(l,x)。又’若是第i分佈線Bclx之變曲點. 冽如有五個時’將其各變曲點SKhx)、SK2,x)、SiP.x)、 Sl(4,x)及 Sl(5,x)之座標設為〔rsl(l,x),zsl(l,x)〕、〔 rsl(2,x), zsi(2,x)〕、〔 rsl(3,x) ’ zsl(3’x)〕、〔 rsl(4,x),zsl(4,x>〕及 〔rsl(5,x),zsl(5,x)〕’則求出此等座標中之z座標, 33 13 Π168Cv(D oCqIcbT2)-1. 5 ...... (18) 31 1311168 Step 16 "Μ 1 y) The appearance of each of the nets in the single crystal of the 夕 夕 夕 榉 早 早 较高 较高 较高 较高 较高Temperature = radius of oxygen precipitate Rp : 冋 low 冋 temperature R, ~Co J Dod t +R, μ t2 19) , again, the above formula (19) can be established as Dv (Cv-Cv& rD〇 c: In the case where the flow rate of the precipitated holes is sufficiently high to fill the volume difference of the C-pre-C mass of the si&fU, the following formula (2〇) can be established: RP = if D v( CV Cve)dt+ Rcr2 ......(2 0) [r β Λ J Step % 17 Here, the above steps 9, 10, 12, 15 and 16 are repeated until the single crystal rod is cooled to a specific value between 800 and 100 (rc, for example 9 〇 (rc or less. The above formulas (9) to (13) and (17) to (20) are obtained by electrical touch calculation. Step 18 Here, the P-band of the above-mentioned south temperature oxygen precipitates is present. The voids and the high-temperature oxygen precipitates are all divided into non-defective areas, and the first equal concentration line HC〗X' is obtained by calculation to divide the defect-free area closer to the single-layer 32 1311168 crystal bottom side than the p-band ( Refer to Figure 9). Step 19 Therefore, the high concentration of oxygen precipitates is divided into the presence of the enthalpy, eight && d 9 long,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, The position of the side of the first 丨 distribution line is: 矽: the equilibrium concentration of the 格子 between the 矽 矽 矽 在 在 在 在 在 m m m m 126 126 126 126 126 126 126 126 126 126 126 126 126 126 126 126 126 126 126 126 126 126 126 126 126 126 126 h Farming degree (a specific value in the range of M2~〇.13%, for example, 〇126%, refers to the towel of the service life (hfetlme), etc. Corresponding to the concentration of the B-band, the B-band means that the condensate of the sputum between the slabs becomes a nucleus and the oxygen precipitates become a high concentration by heat treatment. Step 20 Here, the surface of the first iso-concentration line HC1X is obtained. If the difference between the maximum value of the point and the minimum value of the inflection point of the first distribution line BClxi is, for example, three points of the inflection point of the i-th equal concentration line HC1X, each of the inflection points Q1(l, x), Ql(2) , x) and Ql (3, x) are set to [r Q1(1,x), z Qi(l'x)], [ r Q1(2,x) ' z Q1(2,x)] and [r Q1(3,x),z Q1(3,x)], and find the z coordinate in these coordinates, ie the maximum value ZQi in z Q1(i,x), z Q1(2,x)& z qi(3,x) l, x). 'If it is the curve of the ith distribution line Bclx. For example, if there are five 'transforms each of its points SKhx', SK2, x), SiP.x), Sl(4, x) and Sl(5, The coordinates of x) are set to [rsl(l,x),zsl(l,x)], [rsl(2,x), zsi(2,x)],[ rsl(3,x) ' zsl(3' x)], [rsl(4,x),zsl(4,x>] and [rsl(5,x),zsl(5,x)]' then find the z coordinate in these coordinates, 33 13 Π168

Zs!(l’x)、Zsl(2,x)、zsi(3,x)、Zsi 科繼之,求出第巧濃度心=之最小 大值ZQ1(1,X)與第i分佈線Bdx X變曲點之最 之差ΔΖι(χ)。 明點之取小值Zsi(3,x) 步驟21 在此’將單結晶製造條件之參數 至步驟20之操作;再 2 2’進行步驟2 步-至步㈣之操二進行 為心止,反覆實行上述步称2至步·;H條伟件之參數改 第1等濃度線0之知作後’求出 之變曲點之最小值之差=與第1分佈線吨Zs!(l'x), Zsl(2,x), zsi(3,x), Zsi, followed by the minimum large value ZQ1(1,X) and the ith distribution line Bdx The difference between the X points of change is ΔΖι(χ). The small value Zsi (3, x) of the point is taken. Step 21 Here, the parameters of the single crystal manufacturing condition are changed to the operation of the step 20; then the 2 2' step 2 step-to the step (4) is performed. Repeat the above-mentioned step 2 to step ·; the parameters of the H-pieces are changed to the first-order concentration line 0, and the difference between the minimum values of the calculated points is determined = the same as the first distribution line

Zl(X)為最大之單結晶製造條件。㈣算求出此差△ 度之I::::::之:f之預測計算及修正提拉速 擬方法之步驟卜心同步驟1〜17係與上述缺陷模 步驟18, 溫氣劃分上述高溫氧析*物存在之Ρ帶與空孔及--氧析出物皆無存在之益缺V興:孔及向 :⑷一,以劃分無缺陷領域比;等 側之位置(參照第12圖)。 帶更罪k單,”頂部 步驟19, 同時S算畫=濃度氧析出物存在之B帶及無缺陷領域, 帶更靠近單社曰庇2分佈線BC2x,以劃分無缺陷領域比B 近早、、、。晶底部側之位置。上述第2分佈線係指格子 34 1311168 7矽濃度對矽融點時之格子間矽之平衡濃度在〇12〜〇 i3 %之範圍之特定值,例如〇.126%之等濃度線。在此所謂格 子間矽濃度0.12〜0,13%之範圍之特定值,例如〇126%, 係扣、.星由貫行模擬所得之格子間石夕濃度之中,從使用壽命 之標圖等觀察之對應於B帶之濃度。 步驟20' 在此,求出第2等濃度線HC2X之變曲點之最小值與第 2#分佈線BC2x之變曲點之最大值之差^22〇〇。若是第2等 濃度線HC2X之變曲點例如有三個時,將其各變曲點 ’X)、Q2(2,x)及 Q2(3,x)之座標設為〔r Q2(l,x),z Q2〇,X)〕、〔rQ2(2,x),zQ2(2,x)〕及〔rQ2(3,x),Zq2(3,x)〕, 並求出此等座標中之Z座標,即z Q2( 1,x)、z Q2(2,X)及z Q2(3 ,x) 中之最大值z Q2(1,X)。又,若是第2分佈線BC2x之變曲點 冽如有五個時,將其各變曲點S2(1,x)、S2(2,x)、s2(3,x)、 S2(4,x)及S2(5,x)之座標設為Zl (X) is the largest single crystal production condition. (4) Calculating the I:::::: of the difference Δ degree: the step of predicting the calculation of f and the step of correcting the pull-up method, and the steps 1 to 17 and the above-mentioned defect mode step 18, The high temperature oxidization of the material exists in the presence of sputum and pores and - oxygen precipitates are not present. V. Hing: hole and direction: (4) one to divide the defect-free area ratio; the position of the isolateral side (refer to Figure 12) . Take the more sinful k single," top step 19, while S counts = concentration of oxygen precipitates in the presence of the B-band and non-defective areas, with the band closer to the single community 2 2 distribution line BC2x, to divide the defect-free field than B The position of the bottom side of the crystal. The second distribution line refers to a specific value of the equilibrium concentration of the lattice between the grids at the concentration of the grid 34 1311168 7矽, in the range of 〇12~〇i3%, for example, 〇 .126% of the concentration line. Here, the specific value of the range of the inter-column enthalpy concentration is 0.12~0, 13%, for example, 〇126%, the buckle, the star is calculated by the inter-grid concentration of the grid. The concentration corresponding to the B band is observed from the life chart of the service life, etc. Step 20' Here, the minimum value of the inflection point of the second equal concentration line HC2X and the inflection point of the 2# distribution line BC2x are obtained. The difference between the maximum values is 22. If there are three inflection points of the second equal concentration line HC2X, for example, the coordinates of each of the inflection points 'X), Q2 (2, x) and Q2 (3, x) Set to [r Q2(l,x),z Q2〇,X)], [rQ2(2,x),zQ2(2,x)] and [rQ2(3,x),Zq2(3,x)] And find the Z coordinate in these coordinates, z Q2( 1,x), z Q2(2,X) and z Q2(3 ,x) the maximum value z Q2(1,X). Also, if the second distribution line BC2x is inflection point, if any In five cases, the coordinates of each of the transformation points S2 (1, x), S2 (2, x), s2 (3, x), S2 (4, x), and S2 (5, x) are set.

Zs2(2,X)〕、〔rs2(3,x),zS2(3,x)〕、U2(4,x),Zs2(4,x)〕及 〔rs2(5’x) ’ Zs2(5,x)〕,則求出此等座標中之z座標, ZS2(l’x)、ZS2(2,X)、Zs2(3,x)、Zs2(4,x)及 Zs2(5,x)^ 之最小 值 Zs2(3,x) 〇 步驟21' 在此將矽單結晶製造條件之參數改為P2,進行步驟2 17及步驟18〜2G之操作;再將單結製造條件之參數改 為P3,進行步驟2〜17及步驟18,〜2〇,之操作,直至將 結晶製造條件之參數改為PN止,反覆實行上述操作後,求 35 1311168 ^等;辰度線HC2X之變曲點之最小值與第2分佈線BC2X 之變曲點之最大值之差ΔΖ2(χ),進而依計算求出此差△ Z2(x)為最大之單結晶製造條件。 羡配合比較例詳細說明本發明之實施例於下。 貫施例1 據式(A)控制提拉速度及拉起所定直徑之梦單結晶 棒。 比較例1 根據下式(C)拉起所定直徑之矽單結晶棒。此時設定 PID常數作為頂部以下之提拉所用。Zs2(2,X)], [rs2(3,x),zS2(3,x)], U2(4,x),Zs2(4,x)] and [rs2(5'x) ' Zs2(5 , x)], find the z coordinate in these coordinates, ZS2 (l'x), ZS2 (2, X), Zs2 (3, x), Zs2 (4, x) and Zs2 (5, x) ^The minimum value Zs2(3,x) 〇Step 21' Here, the parameter of the single crystal manufacturing condition is changed to P2, and the operation of step 2 17 and steps 18~2G is performed; and the parameters of the single junction manufacturing condition are changed to P3, carry out steps 2 to 17 and step 18, ~2〇, the operation until the parameter of the crystallization manufacturing condition is changed to PN, and after repeatedly performing the above operation, seek 35 1311168 ^, etc.; the curvature point of the brightness line HC2X The difference between the minimum value and the maximum value of the inflection point of the second distribution line BC2X is ΔΖ2 (χ), and the single crystal production condition in which the difference ΔZ2(x) is the largest is calculated by calculation. The embodiment of the present invention will be described in detail below with reference to comparative examples. Example 1 According to the formula (A), the pulling speed is controlled and the dream single crystal rod of the predetermined diameter is pulled up. Comparative Example 1 A single crystal rod of a predetermined diameter was pulled up according to the following formula (C). At this time, set the PID constant as the pull below the top.

Vn=Vs+P (dn+IZ{(dn.1 + dn)/2}x Δί +D {(dn—dn·,)/ △ t} χ exp(-t/T)〕.........(c) 式(C)中與上述式(A)之記號同一記號表示同一項。 比較例2 根據上式(C)拉起所定直徑之矽單結晶棒。此時設定 PID常數作為頂部之提拉所用。 〈比較試驗及評估〉 第13〜15圖分別表示實施例丨及比較例丨及2之實際 提拉速度對設定提拉速度之變化情形。 # — ^第13〜15圖可知’依比較例1時,實際提拉速度與 «又定提拉速度成一致所需拉起長度較長,約麵(參照第 )比較例2時,因頂部以下之速度變動大,致使品質 :士,差(參照第15圖);實施例i時,實際提拉速度要與 提拉速度&amp;致所需拉起長度明顯較短,約1〇〇腦(參 36 1311168 照第13圖)。 實施例2 如第16(a)圖所示,將所定直徑之石夕單結晶棒培育至所 疋寺間並預測其後之品質。此例為培育至至所定時間時 之提拉結晶長為51 〇圓。在此考慮所定時間之提拉結晶長 度及到該時之實測提拉速度設定檔(pr〇file),而假定繼後該 矽單結晶棒以設定提拉速度完結提拉操作。在此等條件下 實仃培育至所定時間止之結晶部份之品質預測計算。在此 例中,由計算結果看到提拉速度有變化之部分,即自提拉 長度496 mm至510腿止之範圍被預測為可發生L/DL領域。 ’根據此預測結果,為了減少被預測到之L/DL發生領 域,即不良部,一邊修正設定提拉速度,一邊預測不良部 消失之狀況。其結果示於16〇3)圖。於第16(a)&amp;(b)圖中, PI領域為格子間石夕型_陷佔優勢而無格子間石夕凝聚造成 之缺陷⑽以領域。又,pv領域為空孔型點缺陷佔優勢 而無空孔凝聚造成之缺陷之領域。 由於此例為預測會發±L/DL,故將設定提拉速度往上 修正。經此修正,設定提拉速度變快,實際之提拉速度亦 配合該速度加快。結果,回到不發生如l/dl等不良部之設 如上述’於結晶棒之提拉中併行品質預測計算,便可快 速的使貫測之提拉速度與;4 π $ # &gt; &lt; + 生不良部之設定提拉速度趨 一致減低不良部份。甚且,於钍s % 口日日成長後根據全結晶長之 品質預測結果得以決定結晶 曰之切斷位置及品質確認位置。 37 1311168 由上述可知,本發明因對矽單結晶棒之提拉速度加以 控制使矽單結晶棒達到目標直徑之控制方法及控制加熱器 溫度使矽單結晶棒達到目標直徑之控制方法中,分別適用 以複數之階段改變PID常數之PID控制法,因而能在矽單 結晶棒之頂部之提拉時,將PID控制之PID常數設定較大 值而優先的實行矽單結晶棒之直徑控制之控制,同時在矽 2結晶棒之頂部以下之提拉時,階段的減小PID控制之PID 吊數而優先的貫行石夕單結晶棒之提拉速度控制之控制。結 果…矽單結Ba棒之頂部之提拉時,對直徑偏差之提拉速度 ^補正量變大,可快速的使頂部之直徑變動安定,因而矽 單、·Ό Ba棒之頂部以下之提拉時,對直徑偏差之提拉速度之 補正量變小,故可將頂部以下之直徑變動抑制於最小。 又在上述之對石夕單結晶棒之提拉速度加以控制使石夕 單結晶棒達到目標直徑之j D控制方法中組入矽單結晶棒 之目私直控及實測直控之直徑偏差直接回饋之方法及上述 直徑偏差之變化量作為偏差直接回饋現在之提拉速度之方 法,即可在矽單結晶棒之頂部以下之提拉時,將向前次之 單結晶棒之提拉速度實行PID控制回饋時之提拉速度作為 基準,以矽單結晶棒之目標直徑及實測直徑之前次之直徑 偏差減去矽單結晶棒之目標直徑及實測直徑之此次之直徑 偏差之變化量作為偏差補正前次之提拉速度, 部以下之提拉時之石夕單結晶棒之提拉速度之變動门抑制頂 又,將矽單結晶棒之目標直徑與實測直徑之直徑偏差 的變化量作為偏差回饋至矽單結晶之提拉速度時,如對上 38 1311168 4拉速:實订PID控制,使其不超越對現在之提拉速度 之補正之取大變動幅度’則切單結晶棒之提拉時,若超 越對現在之提拉速度之補正之最大變動幅度之場合,便可 Ζ = Ϊ關於最大變動幅度而將⑪單結晶棒之提拉速度 另外’在@早結晶棒之實際提拉中’輕行由提拉速 度貝測設定檔(pr〇file)及設定提拉速度實行石夕單結晶棒之 品質預測計算,即可糊是否會切單結晶棒發生不良 部,若預測到會發生不良部,則算出石夕單結晶棒之修正提 ,速度及修正加熱器溫度而將其回饋至設定提拉速度及設 定加熱器溫度。如此即可減少或阻止石夕單結晶棒之 的發生。 1 ^又,上述之矽單結晶棒之品質之預測計算方法及修正 提拉速度之求算方法’不但可用以求出石夕單結晶棒之内部 溫度’且能藉求算在冷卻過程之料結晶棒内之溫度分 * 佈,即考慮由矽融液脫離之矽單結晶棒之冷卻過程之矽單 \ 結晶棒之徐冷及急冷效果’用電腦求出石夕單結晶棒内之缺鲁 陷之濃度分佈及大小(size)分佈。歡,根據計算求出石夕單 =晶棒内之第i等濃度線及第!分佈線,再根據計算求出 第1等濃度線之變曲點之最大值及第丨分佈線之變曲點之 取小值之差後,改變上述單結晶棒之製造條件之參數,如 上述同樣求出第i等濃度線之變曲點之最大值及第i分佈 線之變曲點之最小值之差。進而求出上述第丨等濃.度線之 - 變曲點之最大值及第〗分佈線之變曲點之最小值之差會為 39 1311168 最大之矽單結晶棒之製造條件。如此g处 出在矽單結晶棒之提拉方向及半徑方^此依6十算正確的求 擴大之矽單結晶棒之製造條件。 D均無缺陷領域最為 再者,計算求出矽單結晶棒内 分佈線,並求出第2等濃度線之變==度線… 佈線之變曲點之最小值之差之後,改變f 2及第2分 製造條件之參數,如同上述依計算求出第;二:;:棒之Vn=Vs+P (dn+IZ{(dn.1 + dn)/2}x Δί +D {(dn—dn·,)/ Δ t} χ exp(-t/T)]..... (c) The same symbol as in the above formula (A) in the formula (C) indicates the same item. Comparative Example 2 A single crystal rod of a predetermined diameter is pulled up according to the above formula (C). At this time, the PID constant is set. As a top pull. <Comparative Test and Evaluation> Figures 13 to 15 show the change of the actual pull speed of the example and the comparative example 丨 and 2 to the set pull speed. # — ^第13〜15 As can be seen from the figure, in the case of Comparative Example 1, the actual pulling speed is longer than the predetermined pulling speed, and the pulling length is long. When the surface is compared with the comparative example 2, the speed fluctuation below the top is large. Quality: gentleman, poor (refer to Figure 15); in the case of example i, the actual pulling speed and the pulling speed & required to pull up the length is significantly shorter, about 1 camphor (see 36 1311168 according to the 13th Fig. 2 Example 2 As shown in Fig. 16(a), the stone-shaped single crystal rod of the determined diameter is cultivated between the temples and predicted to be followed by the quality. This example is to raise the seedlings until the time is fixed. The crystal length is 51 〇 round. Considering the length of the pulling crystallization of the determined time and the measured pull speed profile (pr〇file) at that time, it is assumed that the single crystallizing rod is then pulled at the set pulling speed to complete the pulling operation.品质 仃 品质 品质 品质 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 仃 品质 品质 品质 品质 品质 品质 品质 品质 品质 品质 品质The L/DL field is generated. 'According to this prediction result, in order to reduce the L/DL occurrence area that is predicted, that is, the defective part, the situation in which the defective part disappears is corrected while the lifting speed is set. The result is shown in 16〇3. ) Figure. In the 16th (a) &amp; (b) diagrams, the PI field is a field in which the defects (10) caused by the lattice-type shi _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Further, the pv field is an area in which a hole type defect is dominant and there is no defect caused by hole agglomeration. Since this example will generate ±L/DL for the prediction, the set pull speed will be corrected upward. With this correction, the pulling speed is set to be faster, and the actual pulling speed is also increased in conjunction with the speed. As a result, returning to the non-defective part such as l/dl, such as the above-mentioned parallel quality prediction calculation in the pulling of the crystal rod, the pulling speed of the continuous measurement can be quickly made; 4 π $ # &gt;&lt;; + The setting of the bad part of the birth is consistent with the reduction of the bad part. In addition, after the growth of the 钍s% port, the quality of the crystallization is determined based on the results of the quality prediction of the crystallization. 37 1311168 It can be seen from the above that the present invention controls the pulling speed of the single crystal rod to control the single crystal rod to reach the target diameter and the control method for controlling the heater temperature to make the single crystal rod reach the target diameter. It is applicable to the PID control method of changing the PID constant at the stage of the plural, so that the PID constant of the PID control can be set to a larger value when the top of the single crystal rod is pulled, and the control of the diameter control of the single crystal rod is preferentially carried out. At the same time, when lifting at the top of the 矽2 crystallizing rod, the phase of the PID control of the number of PID hangs is reduced, and the control of the pulling speed control of the singular single crystal rod is given priority. As a result, when the top of the Ba-bar is lifted, the pulling speed of the diameter deviation is increased, and the diameter of the top is made to be stable, so that the diameter of the top of the bar can be stabilized. In this case, since the correction amount of the pulling speed of the diameter deviation becomes small, the diameter variation below the top can be suppressed to the minimum. In addition, in the above-mentioned j D control method for controlling the pulling speed of the Shixi single crystal rod, the diameter deviation of the direct control of the single crystal rod and the direct measurement of the direct measurement are directly The method of feedback and the change amount of the above-mentioned diameter deviation as a method of directly feeding back the current pulling speed as a deviation can perform the pulling speed of the previous single crystal rod when pulling at the top of the single crystal rod. The pulling speed of the PID control feedback is used as a reference, and the deviation of the diameter of the target diameter of the single crystal rod and the diameter of the measured diameter minus the diameter of the target diameter of the single crystal rod and the diameter of the measured diameter is used as the deviation. Correcting the lifting speed of the previous one, and the variation of the pulling speed of the single crystal rod of the Shixi single crystal rod under the lower part of the lifting, and the variation of the diameter deviation between the target diameter of the single crystal rod and the measured diameter as the deviation When feeding back to the pulling speed of the single crystal, for example, the speed of the upper 38 1311168 4: the actual PID control, so that it does not exceed the correction of the current lifting speed When the single crystal rod is pulled, if it exceeds the maximum fluctuation range of the current lifting speed correction, 提 = Ϊ about the maximum fluctuation range and the pulling speed of the 11 single crystal rod is additionally 'pre-crystallized at @ In the actual lifting of the rod, the light weight is calculated from the pr〇file and the set pulling speed. The quality prediction of the stone singular crystal rod is calculated. If it is predicted that a defective part will occur, calculate the correction of the Shixi single crystal rod, speed and correct the heater temperature, and feed it back to the set pulling speed and set the heater temperature. This can reduce or prevent the occurrence of the stone single crystal rod. 1 ^ In addition, the above-mentioned method for predicting the quality of single crystal rods and the method for calculating the lifting speed are not only used to determine the internal temperature of the stone core, but can be calculated in the cooling process. The temperature in the crystal rod is divided into * cloth, that is, considering the cooling process of the single crystal rod which is separated from the crucible melt, the single crystal of the crystal rod and the quenching effect of the crystal rod are determined by the computer. Concentration distribution and size distribution. Huan, according to the calculation to find Shi Xidan = the i-th concentration line in the ingot and the first! After dividing the wiring, and calculating the difference between the maximum value of the inflection point of the first equal concentration line and the small value of the inflection point of the second distribution line, the parameters of the manufacturing conditions of the single crystal rod are changed, as described above. Similarly, the difference between the maximum value of the inflection point of the i-th concentration line and the minimum value of the inflection point of the i-th distribution line is obtained. Further, the difference between the maximum value of the inflection point and the minimum value of the inflection point of the first distribution line of the second confinement line is obtained as the manufacturing condition of the maximum single crystal rod of 39 1311168. Such a g is produced in the pulling direction and radius of the single crystal rod, and the manufacturing conditions of the single crystal rod are expanded according to the accuracy of the sixty-six. D is the most no defect field, calculate the distribution line in the single crystal rod, and find the difference of the second equal concentration line == degree line... the difference between the minimum points of the bending points of the wiring, change f 2 And the parameters of the second sub-manufacturing conditions, as described above by the calculation; second:;: rod

曲點之最大值及第2分佈線之變 ::線之變 , 叫點之取小值之差,進而 鐵上迹第1料度線之變曲點之最大值與第〗分佈線之 1曲=之最小值之差會為最大之料結晶棒之製造條件亦 可獲彳于上述同樣之效果。 由上述可知,本發明之矽單結晶棒之製造方法可利用 在對切克勞斯基法提拉(拉起)之矽單結晶棒施予pm控 制以抑制矽單結晶棒之直徑變動,製取不具微細缺陷之 晶圓。The maximum value of the curved point and the change of the second distribution line: the change of the line, the difference between the small points of the point, and the maximum value of the curve of the first line of the iron track and the first line of the distribution line The difference between the minimum value of the curve = the maximum production condition of the crystallized rod can also be obtained by the same effect as described above. From the above, it can be seen that the method for producing a single crystal rod of the present invention can be applied to a single crystal rod of a Czochralski method (pull up) to control the diameter variation of the single crystal rod. Take wafers with no micro defects.

40 1311168 【圖式簡單說明】 事置之第丄圖為本發明第1及第2實施形態之矽單結晶棒提拉 衷置之縱剖面圖; 叫 第2圖為提拉發軍結晶棒時之控制操作的方塊圖; 第3圖為本發明第3實施形態之矽單結晶棒 制操作的方塊圖; 曰曰棒k拉時之控 圖; .苐4圖為預測該石夕單結晶棒之品質之前段操作的流程 第5圖為預測該梦單社a摄夕σ所二4 圖; . 早、、,°日日棒之。口質之後段操作的流.程 圖; 第6圖為矽單結晶棒之缺陷模擬方法之 第一段的流程 圖; .第7圖為矽單結晶棒之缺陷模擬方法之第 二段的流程 圖; 圖; 第8圖為矽單結晶棒之缺陷模擬方法 第9圖為矽單結晶棒之缺陷模擬方法 之第三段的流程 之第四段的流程 五段的流輕 .第10®為石夕單結晶棒之缺陷模擬方法之 圖, 結晶之提拉機 第U圖為將矽融液作成網目構造之矽 之要部剖面圖; 早 第12圖為表示改變該矽單結曰 晶内之格子間矽及空孔之分佈說明圖;曰拉速度時之矽單結 41 1311168 之變化第線 13圖圖為實施例…單結晶棒之提拉長與提拉速度 之變:二圖為比較例1…結晶棒之提拉長與提拉速度 之變㈣,繼拉速度發生時第,施例2之”結晶棒預測到有不良部 狀態圖》 …器皿度以減少不良部之發生之 【符號說明】 10......石夕單結晶提拉裝置 π — 12a •主室 内層容器 12 13......矽融液 16......驅動裝置 18......保溫筒 21......提拉軸 23......種晶夾持具 12b - -掛堝 外層容器 14......軸 17......加熱器 19 提拉室 22------種晶 24......矽單結晶棒 4240 1311168 [Comprehensive description of the drawings] The second diagram of the present invention is a longitudinal sectional view of the single crystal rod of the first and second embodiments of the present invention; FIG. 3 is a block diagram showing the operation of the single crystal rod in the third embodiment of the present invention; the control chart when the rod is pulled; and the figure 4 is for predicting the single crystal rod of the stone The quality of the previous paragraph of the process of the fifth figure is to predict the dream of a single photo of the sigma of the sigma 2; 4; early,,, ° day and day. Flow chart of the operation of the latter part of the mouth; Figure 6 is a flow chart of the first stage of the defect simulation method of the single crystal rod; Figure 7 is the flow of the second stage of the defect simulation method of the single crystal rod Fig. 8 is a simulation method of the defect of a single crystal rod. Fig. 9 is a flow of the fifth stage of the fourth stage of the process of the defect simulation method of the single crystal rod. The 10th is The figure of the simulation method of the defect of the stone singular crystal rod, the U drawing of the crystal pulling machine is a sectional view of the main part of the ramming liquid which is made into the mesh structure; the early 12th picture shows the change of the 矽 single knot 曰 crystal The distribution diagram of the 矽 矽 空 空 曰 曰 曰 曰 曰 曰 曰 曰 曰 曰 曰 曰 曰 曰 曰 曰 曰 曰 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 Comparative Example 1: The elongation of the crystal rod and the change of the pulling speed (4), when the pulling speed occurs, the "crystal stick of the example 2 is predicted to have a bad state diagram" ... the degree of the dish is reduced to reduce the occurrence of the defective portion. [Symbol description] 10...Shixi single crystal pulling device π-12a • Main room Container 12 13 ... 矽 melt 16 ... drive device 18 ... insulation tube 21 ... pull shaft 23 ... seed clip Holder 12b - - 埚 埚 outer container 14 ... shaft 17 ... heater 19 pull chamber 22 ------ seed crystal 24 ... 矽 single crystal rod 42

Claims (1)

1311168__ 公告本 ---- (V '' 拾、申請專利範圍: 【申請專利範圍】 1·一種矽單結晶之製造方法,係從藉由加熱器(17)融解 之矽融液提拉矽單結晶棒(24),在此提拉中按所定之時間間 隔檢測該矽單結晶棒(24)之直徑變化,而將該檢測資料回饋 至上述矽單結晶棒之提拉速度及上述加熱器(17)溫度而控 制上述發卓結晶棒之直控之方法,其特徵包括: 對於控制上述之矽單結晶棒(24)達到目標直徑之上述 矽單結晶棒(24)之提拉速度之控制方法,及控制上述矽單結 晶棒達到目標直徑之上述加熱器(17)溫度之控制方法,分別 多階段的運用改變PID常數之PID控制而 (1) 在矽單結晶棒(24)之頂部之提拉時,將上述piD控 制之PID常數設定較大值以增大矽單結晶棒(24)之對直徑 偏差之提拉速度之補正量而優先的實行矽單結晶棒(24)之 直徑控制之控制;及 (2) 在矽單結晶棒(24)之頂部以下之提拉時,配合設定 提拉速度將上述PID控制之PID常數階段的變小以減小矽 單結晶棒(24)之對直徑偏差之提拉速度之補正量而優先的 實行梦單結晶棒(24)之提拉速度控制之控制。 2· —種石夕單結晶之製造方法,係從藉由加熱器(丨7)融解 之矽融液提拉矽單結晶棒(24),在此提拉中按所定之時間間 隔檢測該矽單結晶棒(24)之直徑變化,而將該檢測資料回饋 至上述矽單結晶棒之提拉速度及上述加熱器(17)溫度而控 43 1311168 制上述石夕單結晶棒之直徑之方法,係以: 對於控制上述之矽單結晶棒(24)達到目標直徑之上 述石夕單結晶棒(24)之提拉速度之PID控制方法中,組入一將 上述矽單結晶棒(24)之目標直徑與實測直徑之直徑偏差直 接回饋之方法及一將上述直徑偏差之變化量作為偏差回饋 至現在之提拉速度之方法為其特徵者。 3 · —種矽單結晶之製造方法,係從藉由加熱器(丨7)融解 之石夕融液提拉石夕單結晶棒(24),在此提拉中按所定之時間間 隔檢測該矽單結晶棒(24)之直徑變化,而將該檢測資料回饋 至上述矽單結晶棒之提拉速度及上述加熱器(17)溫度而控 制上述矽單結晶棒之直徑之方法,係以: 於將上述矽單結晶棒(24)之目標直徑與實測直徑之直 徑偏差的變化量作為偏差,回饋至上述矽單結晶棒之提拉 速度時,將上述提拉速度PID控制,使其不超越對現在之 提拉速度之補正之最大變動幅度為其特徵者。 4.如申請專利範圍第1〜3項之任一項之矽單結晶之製 造方法,其中在矽單結晶棒(24)之提拉時,使用提拉開始至 所疋時間止之提拉速度實測設定檔(pr〇me)及提拉開始至 提拉終了止之設定提拉速度,並併行上述矽單結晶棒之品 質預測計算,以預測上述矽單結晶棒是否會發生不良部, 而預測到會發生上述不良部時,則算去修正該不良部所要 之上述矽單結晶棒之修正提拉速度及修正加熱器溫度,而 44 1311168 正加熱器溫度回饋至上述設定提 將該等修正提拉速度及修 拉速度及設定加熱器溫度 。5.如申請專利範圍第4項之矽單結晶之製造方法,其 中石夕單結晶棒(24)之品質預測計算及修正提拉速度之計算 係用含有下述步驟: 、、⑷於參數Pl之石夕單結晶棒㈣之製造條件,考慮石夕融 液(13)之對流’求算從上述㈣液(13)成長之上述碎單結晶 棒内之溫度分佈之步驟; (b)藉求出在冷卻過程之上述矽單結晶棒内之溫度分 佈,預測上述♦單結晶棒内之空隙(vGid)及高溫氧析出物之 濃度分佈及大小(sjze)分佈之步驟; ⑷於計算求出上料單結晶棒内之第1等濃度線及第 分佈線之後,計算求出第1等濃度線之變曲點之最大值與 第1分佈線之變曲點之最小值之差之步驟; ⑷將上述⑪單結晶棒之製造條件之參數由&amp;依次改變 至pN,依計算求出上述第!等濃度線之變曲點之最大值與 上述第1分佈線之變曲點之最小值之差之步驟;及 (e)求出第1等濃度線之變曲點之最大值與上述第1分 佈線之變曲點之最小值之差可為最大時之上述發單結晶棒 之製造條件之步驟;之電腦使上述石夕單結晶棒之無缺陷領 域可最大化之缺陷模擬方法實行者。 6·如申請專利範圍第4項之矽單結晶之製造方法,其t 45 1311168 該矽單結晶棒(24)之品質預測計算及修正提拉速度之計算 係用含有下述步騾: (a)於參數P〗之矽單結晶棒(24)之製造條件,考慮矽融 液(13)之對流,求算從上述㈣液(13)成長之上料翠結晶 棒内之溫度分佈之步驟; (b)藉求出在冷卻過程之上述矽單結晶棒内之溫度分 佈,預測上料單結晶棒内之空隙(vGid)及高溫氧析出物之 濃度分佈及粒度(size)分佈之步驟; 2八計算巧上述料結晶棒内之第2等濃度線及角 第' 八二後:算求出第2等濃度線之變曲點之最大值1 第2刀佈線之變曲點之最小值之差之步驟; ⑷將上财單結晶棒之製造條件之參數 至PN,依計算求出上诫坌?笙.曲麻祕 2伙人汉驾 上述第2八佑綠11第2也辰度線之變曲點之最大㈣ 上达第刀佈線之變曲點之最小值之差之步驟 ⑷求出第2等濃度線之變曲點之最大值 佈線之變曲點之最小值之差A 、上这第2刀 之製造條件之牛醉為最大時之上述石夕單結晶相 之裊以條件之步驟;之電腦使上曰 域可最大化之缺陷模擬方法實行者广,棒之無編 461311168__ Announcement---- (V '' Pickup, patent application scope: [Scope of application for patents] 1. A method for manufacturing single crystals, which is carried out from the melt of the melt by the heater (17) a crystallization rod (24) for detecting a change in diameter of the bismuth single crystal rod (24) at a predetermined time interval in the pulling, and feeding the detection data back to the pulling speed of the bismuth single crystal rod and the heater ( 17) A method for controlling the direct control of the above-mentioned crystallizing rod, characterized in that: a method for controlling the pulling speed of the above-mentioned single crystal rod (24) for controlling the above-mentioned single crystal rod (24) to reach a target diameter And controlling the temperature of the heater (17) at which the single crystal rod reaches the target diameter, and the multi-stage application is used to change the PID control of the PID constant (1) on the top of the single crystal rod (24) When pulling, the PID constant of the above piD control is set to a larger value to increase the correction amount of the pulling speed of the diameter deviation of the single crystal rod (24), and the diameter control of the single crystal rod (24) is preferentially performed. Control; and (2) in the order When the top of the ingot (24) is pulled below the tip, the PID constant phase of the PID control is reduced to reduce the correction amount of the diameter deviation of the single crystal rod (24). And the control of the pulling speed control of the dream single crystal rod (24) is preferentially carried out. 2·—The manufacturing method of the single crystal of the stone is extracted from the melt of the melt by the heater (丨7) a crystallization rod (24) for detecting a change in diameter of the bismuth single crystal rod (24) at a predetermined time interval in the pulling, and feeding the detection data back to the pulling speed of the bismuth single crystal rod and the heater ( 17) Temperature control 43 1311168 The method for making the diameter of the above-mentioned single crystal rod is: for controlling the pulling speed of the above-mentioned single crystal rod (24) of the single crystal rod (24) reaching the target diameter In the PID control method, a method of directly feeding back the diameter deviation between the target diameter of the single crystal rod (24) and the measured diameter and a variation of the diameter deviation as a deviation is fed back to the current pulling speed. The method is its characteristic. 3 · A method for producing a single crystal of sputum is to extract a stone single crystal rod (24) from a stone melt melted by a heater (丨7), and the strip is detected at a predetermined time interval in the pulling The method for controlling the diameter of the single crystal rod (24) and feeding the test data to the pulling speed of the single crystal rod and the temperature of the heater (17) to control the diameter of the single crystal rod is as follows: When the amount of change in the deviation between the target diameter of the single crystal rod (24) and the diameter of the measured diameter is used as a deviation, and the feedback speed of the single crystal rod is fed back, the pulling speed PID is controlled so as not to exceed The greatest change in the current correction speed is characteristic. 4. The method for producing a single crystal according to any one of claims 1 to 3, wherein in the pulling of the single crystal rod (24), the pulling speed from the start of the pulling to the time of the winding is used. The measured setting file (pr〇me) and the lifting speed from the start of the lifting to the end of the lifting, and parallel calculation of the quality of the above-mentioned single crystal rod to predict whether the above-mentioned single crystal rod will have a bad part, and the prediction When the above-mentioned defective part occurs, the corrected pulling speed of the single crystal rod and the corrected heater temperature of the defective part are corrected, and the 44 1311168 positive heater temperature is fed back to the above setting. Pull speed and pull speed and set heater temperature. 5. The method for manufacturing a single crystal according to item 4 of the patent application scope, wherein the calculation of the quality prediction of the Shixi single crystal rod (24) and the calculation of the modified pulling speed include the following steps: , (4) in the parameter P1 The manufacturing conditions of the stone singular single crystal rod (4), considering the convection of Shi Xirong liquid (13), the step of calculating the temperature distribution in the above-mentioned broken single crystal rod grown from the above (4) liquid (13); (b) borrowing The temperature distribution in the above-mentioned single crystal rod in the cooling process is predicted, and the steps of the distribution (vGid) in the single crystal rod and the concentration distribution and size (sjze) distribution of the high temperature oxygen precipitate are predicted; (4) After the first equal concentration line and the first distribution line in the single crystal rod, the step of determining the difference between the maximum value of the inflection point of the first isocratic line and the minimum value of the inflection point of the first distribution line is calculated; (4) The parameters of the manufacturing conditions of the above 11 single crystal rods are sequentially changed from &amp; to pN, and the above-mentioned first is calculated by calculation! a step of the difference between the maximum value of the inflection point of the iso-concentration line and the minimum value of the inflection point of the first distribution line; and (e) determining the maximum value of the inflection point of the first iso-concentration line and the first The difference between the minimum values of the inflection points of the sub-wirings may be the step of manufacturing the above-mentioned billing crystal rods at the maximum; the computer is the one that maximizes the defect-free simulation method of the above-mentioned stone-like single crystal rods. 6. If the manufacturing method of the single crystal of the fourth application of the patent scope is t 45 1311168, the quality prediction calculation and the calculation of the modified pulling speed of the single crystal rod (24) have the following steps: In the parameter P, the manufacturing condition of the single crystal rod (24), considering the convection of the mash liquid (13), the step of calculating the temperature distribution in the crystallization rod from the above (4) liquid (13); (b) a step of predicting a void distribution (vGid) in the single crystal rod and a concentration distribution and a size distribution of the high temperature oxygen precipitate by determining a temperature distribution in the above-mentioned single crystal rod in the cooling process; 8. Calculate the second equal concentration line and the angle in the above-mentioned material crystal rod. After the 'eighth two: calculate the maximum value of the inflection point of the second equal concentration line. 1 The minimum value of the inflection point of the second knife wiring. Step of the difference; (4) Put the parameters of the manufacturing conditions of the upper crystal crystallization rod to PN, and calculate the upper 依 according to the calculation?笙.曲麻秘 2, the first of the 2nd eight-year-old green, the second, the second, the second point of the curve, the maximum of the change point (four), the step of the difference between the minimum of the change point of the first knife wiring (4) The difference between the minimum value of the inflection point of the wiring of the 2 equal-concentration line, and the difference between the minimum of the inflection point of the wiring of the second knife, and the condition of the above-mentioned condition of the second knife. The computer makes the upper 曰 domain maximize the defect simulation method is widely implemented, the stick is not edited 46
TW92126275A 2003-09-24 2003-09-24 Method for manufacturing silicon monocrystal TWI311168B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW92126275A TWI311168B (en) 2003-09-24 2003-09-24 Method for manufacturing silicon monocrystal

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW92126275A TWI311168B (en) 2003-09-24 2003-09-24 Method for manufacturing silicon monocrystal

Publications (2)

Publication Number Publication Date
TW200512321A TW200512321A (en) 2005-04-01
TWI311168B true TWI311168B (en) 2009-06-21

Family

ID=45072373

Family Applications (1)

Application Number Title Priority Date Filing Date
TW92126275A TWI311168B (en) 2003-09-24 2003-09-24 Method for manufacturing silicon monocrystal

Country Status (1)

Country Link
TW (1) TWI311168B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113463184B (en) * 2021-06-29 2023-05-05 西门子(中国)有限公司 Control method, device and readable medium for lifting speed in crystal growth process

Also Published As

Publication number Publication date
TW200512321A (en) 2005-04-01

Similar Documents

Publication Publication Date Title
JP4808832B2 (en) Method for producing defect-free crystals
US6045610A (en) Methods of manufacturing monocrystalline silicon ingots and wafers by controlling pull rate profiles in a hot zone furnance
US6702892B2 (en) Production device for high-quality silicon single crystals
US6472040B1 (en) Semi-pure and pure monocrystalline silicon ingots and wafers
JP4380537B2 (en) Method for producing silicon single crystal
TWI722480B (en) Method for growing silicon single crystal
JP2001500468A (en) Self-interstitial-dominated silicon with low defect density
JP6604338B2 (en) Silicon single crystal pulling condition calculation program, silicon single crystal hot zone improvement method, and silicon single crystal growth method
JP5120337B2 (en) Silicon single crystal manufacturing method, silicon single crystal temperature estimation method
TW201100595A (en) Methods of making an unsupported article of semiconducting material by controlled undercooling
Razumovskii et al. New generation of Ni-based superalloys designed on the basis of first-principles calculations
TW201012984A (en) Generating a pumping force in a silicon melt by applying a time-varying magnetic field
JP2008105873A (en) Single crystal manufacturing apparatus and single crystal manufacturing method
TW200902774A (en) Method for manufacturing silicon single crystal
TW200815628A (en) Single crystal manufacturing apparatus and method
TWI311168B (en) Method for manufacturing silicon monocrystal
JP2010037114A (en) Method for growing silicon single crystal and method for estimating temperature
Kakimoto Development of crystal growth technique of silicon by the Czochralski method
TWI746494B (en) Method for adjusting the resistivity of a semiconductor ingot during its manufacture
KR101862157B1 (en) Method and apparatus for manufacturing silicon monocrystalline ingot
KR100665683B1 (en) Method for manufacturing silicon single crystal
KR100846632B1 (en) Method of manufacturing silicon single crystal, silicon single crystal ingot and wafer fabricated by the same
JP5509636B2 (en) Defect analysis method of silicon single crystal
JP4154936B2 (en) Single crystal defect-free region simulation method
JP2000016897A (en) Production of high quality silicon single crystal

Legal Events

Date Code Title Description
MK4A Expiration of patent term of an invention patent