JP2014110414A - Method and system for forming thin film - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method by which a thin film having a resistivity distribution close to a targeted resistivity distribution can be easily obtained in forming a thin film on the main surface of a semiconductor substrate.SOLUTION: A main dopant is supplied from both of an inside injector connected to the inner region of a reaction chamber and an outside injector connected to the outer region, and an additional dopant for compensating auto-doping is supplied from the inside injector. A method for forming a thin film includes the steps of: preparing a predictive equation giving a predictive resistivity distribution of an epitaxial thin film on the basis of parameters affecting on the epitaxial thin film (S1); setting a targeted resistivity distribution and other parameter values other than supply amounts of the main dopant and the additional dopant (S2); calculating optimum supply amounts of the main dopant and the additional dopant so that the difference between a predictive resistivity distribution obtained by entering the set values having been set in S2 into the predictive equation and the targeted resistivity distribution having been set in S2 becomes minimum (S3); and forming an epitaxial thin film in an epitaxial growth system according to the optimum supply amounts (S4).

Description

  The present invention relates to a thin film forming method and a thin film forming system for forming a thin film such as a silicon single crystal film on a main surface of a semiconductor substrate such as a silicon wafer by vapor phase growth.

  Conventionally, in an epitaxial growth apparatus (semiconductor thin film growth apparatus) for vapor phase growth (epitaxial growth) of a semiconductor thin film (epitaxial thin film) on the main surface of a semiconductor substrate such as a silicon wafer, the main source gas for vapor phase growth is divided into a plurality of regions. An epitaxial growth apparatus that divides and supplies the reaction chamber is known (see, for example, Patent Document 1). FIG. 2 shows a top view of this type of epitaxial growth apparatus. The semiconductor substrate W is supported by a rotatable susceptor 12, and the reaction chamber 2 is formed airtight so as to surround it. The reaction gas containing the main source gas enters the reaction chamber 2 from the injector 50 divided into a plurality on the left side, and is discharged from the right outlet, leaving an epitaxial thin film on the semiconductor substrate W.

  The main reason why the injector 50 is divided is that if a part of the film thickness is thinner than that of the other part (outer peripheral part or the like), the flow rate of the injector 50 corresponding to that part is larger than that of the other injectors 50. This is because the film thickness can be made uniform by increasing the thickness.

  Another important characteristic of epitaxial thin films is resistance. Usually, the resistance of the epitaxial thin film is adjusted by mixing a small amount of dopant gas into the main raw material gas. The dopant gas is introduced into the reaction chamber 2 from the injector 50 together with the main source gas, decomposes on the main surface of the substrate W, and is taken into the epitaxial thin film.

  At this time, if the dopant gas is uniformly introduced into the main raw material gas, the dopant should be uniformly taken into the thin film to obtain a uniform resistance. However, in reality, a uniform resistance is obtained by a phenomenon called auto-doping in which a part of the dopant remaining in the reaction chamber 2 or part of the dopant contained in the substrate W is released into the gas phase and taken into the thin film. It may not be possible.

  Since the influence of auto-doping is usually large around the wafer, the typical resistance distribution of the epitaxial thin film affected by auto-doping is low in the peripheral part incorporating the additional dopant by auto-doping and high in the central part as shown in FIG. become. Note that the horizontal axis of FIG. 17 indicates the in-plane position of the thin film. Specifically, the center position of the thin film (semiconductor substrate W) is zero, and the distance from the center position to the width direction WD of FIG. ing. The vertical axis in FIG. 17 indicates the resistance value.

  In order to suppress the formation of such a resistance distribution, it is most effective to suppress the generation of the dopant which is the source of auto-doping. For this purpose, a film for suppressing the emission of the dopant is formed on the substrate W. Or the released dopant needs to be purged over time, leading to an increase in cost and a decrease in productivity.

  Therefore, as a simple method, a method is adopted in which the resistance is made uniform by adding an additional amount of dopant equivalent to the autodope in the peripheral portion to the central portion of the wafer which is not easily affected by autodope. FIG. 18 shows the resistance distribution of the thin film by this method.

JP 2002-231641 A

  However, with this method of adding additional dopants, how much additional dopant is added is comparable to autodoping and how much main dopant must be reduced to compensate for the resistance reduced by the additional dopant? Must be determined experimentally through trial and error. Therefore, it takes a long time to obtain a resistance distribution close to the target resistance distribution.

  The present invention has been made in view of the above problems, and provides a thin film forming method and a thin film forming system capable of easily obtaining a thin film having a resistance distribution close to a target resistance distribution when forming a thin film on a main surface of a semiconductor substrate. The issue is to provide.

  The amount of main dopant and additional dopant can be determined only by trial and error, in addition to the amount of main dopant and additional dopant, concentration of main source gas, flow rate, flow rate distribution, substrate temperature, substrate resistance Since the resistance distribution of the thin film obtained is determined by many parameters such as substrate backside treatment, susceptor shape, susceptor rotation speed, etc., the main dopant as the cause is added from the desired resistance distribution as a result. This is because it is practically impossible to create a calculation formula that reversely calculates the amount of dopant supplied. Therefore, the present inventor has given up the creation of a reverse calculation formula, and has arrived at the method of the present invention for predicting the resistance distribution by calculating forward from the cause parameter.

That is, in the thin film forming method according to the present invention, the main source gas is divided into a plurality of regions and supplied to the reaction chamber, and the first dopant supply line and the second dopant supply line are connected to different divided regions. A thin film forming method for forming a thin film on a main surface of a semiconductor substrate using a semiconductor thin film growth apparatus,
A first supply amount that is a supply amount of a dopant supplied from the first dopant supply line to the reaction chamber, and a second supply amount that is a supply amount of a dopant supplied from the second dopant supply line to the reaction chamber. A prediction formula creation process for creating a prediction formula that gives a predicted resistance distribution of the thin film, including parameters that affect the resistance distribution of the thin film including the supply amount;
A difference between an expected resistance distribution given by the prediction formula in which other parameters that are parameters other than the first supply amount and the second supply amount are fixed to a preset setting value, and a preset target resistance distribution of the thin film is A supply amount determining step for determining an optimum supply amount that is a combination of the first supply amount and the second supply amount that are the smallest;
A thin film forming step of forming a thin film on the main surface of the semiconductor substrate by vapor phase growth in the reaction chamber according to the set value in the other parameter and the optimum supply amount;
It is characterized by providing.

  Thus, in the present invention, a prediction formula for predicting the resistance distribution of the thin film is created from the parameter (cause parameter) that affects the resistance distribution of the thin film (prediction formula creating step). When the dopant supplied from the first dopant supply line is the main dopant and the dopant supplied from the second dopant supply line is the additional dopant, the amount of the main dopant (first supply amount) and the additional dopant The amount (second supply amount) of the main dopant and the amount of additional dopant is fixed at a preset value (other parameters) other than the amount of the main dopant and the additional dopant, and the amount of the main dopant and the additional dopant is variously combined. It is determined by selecting a combination that provides an expected resistance distribution closest to the desired resistance distribution (target resistance distribution) from the prediction formula (supply amount determining step).

  In other words, it can be said that the present invention replaces the work that has been trial and error on an actual apparatus with conventional trial and error. Computational trial and error can be performed in an instant using today's computer technology, and the optimal amount of main dopant and additional dopant (optimum supply) can be quickly determined. That is, once the prediction formula is created, the optimum supply amount corresponding to the set values of the other parameters and the target resistance distribution can be quickly obtained thereafter. The optimum supply amount is the supply amount with the smallest difference between the predicted resistance distribution and the target resistance distribution. In the thin film formation process, vapor deposition is performed according to the optimum supply amount, so the dopant supply amount is determined by trial and error. Compared with the conventional method, the thin film having a resistance distribution close to the target resistance distribution can be easily obtained.

  Hereinafter, the present invention will be described in more detail. In order to realize the present invention, it is first necessary to create a prediction formula for resistance distribution in the prediction formula creation step. The resistance distribution is determined by the growth rate of the thin film, the concentration distribution of the intentionally added dopant (main dopant, additional dopant) in the gas phase (hereinafter referred to as air concentration distribution) and the air concentration distribution of the dopant by autodoping. . Therefore, the prediction formula can be a formula using the growth rate, the air concentration distribution of the dopant added intentionally, and the air concentration distribution of the dopant by autodoping as parameters. However, since these parameters are secondary parameters that cannot be directly controlled, a function for determining these secondary parameters from the primary parameters that can be directly controlled is first created.

  In the prediction formula creation process, first, a parameter (first parameter) that affects the growth rate of the thin film is changed to examine the growth rate, and a function (first function) that calculates the growth rate from the first parameter based on the result of the investigation. ) Is created (first creation step). In addition, the parameter (second parameter) affecting the amount of autodoping is changed to check the amount of autodoping, and the function (second value) is used to calculate the air concentration distribution of the dopant by autodoping from the second parameter based on the result of the investigation. Function) is created (second creation step). Furthermore, a function for calculating the flow region from the third parameter based on the result of examining the flow region by changing the parameter (third parameter) that affects the region in which the separately supplied main source gas flows in the reaction chamber (Third function) is created (third creation step). By adding the parameters of the main dopant and the amount of additional dopant (first supply, second supply) to the function of the flow region, the function of the concentration distribution of the intentionally added dopant (fourth function) ) Is obtained (fourth creation step).

  From the parameters (primary parameters) of these functions (first function, second function, and fourth function), the growth rate, the air concentration distribution of the dopant added intentionally, and the air concentration distribution (secondary parameter) by autodoping From the obtained secondary parameters, a prediction formula that gives the dopant concentration distribution of the thin film, that is, the predicted resistance distribution, can be obtained (fifth preparation step). Since the dopant concentration distribution (resistance distribution) of the thin film changes depending on the growth rate, the air concentration distribution of the dopant added intentionally, the air concentration distribution due to auto-doping, and the incorporation rate of the dopant incorporated into the thin film, In the fifth creating step, it is preferable to create a prediction formula in addition to the parameter of the capturing speed.

  Next, the distance between the target resistance distribution and the expected resistance distribution is defined. The simplest definition is a method of selecting several points and summing the difference between the expected resistance and the target resistance by squaring, but other definitions may be used. With the above preparation, it is possible to calculate the expected resistance distribution under an arbitrary condition, and it is possible to calculate how close the distribution is to the target resistance distribution.

  Next, parameters other than the main dopant and additional dopant (other parameters) are fixed to preset values, and the amount of the main dopant and additional dopant (first supply amount, second supply amount) varies. Under the combined conditions, the distance between the target resistance distribution and the predicted resistance distribution can be calculated using the prepared prediction formula, and the optimum supply amount of the main dopant and the additional dopant can be determined by searching for the closest combination. .

  The method of searching for the combination having the closest distance to the target is, for example, an n-level range in which the supply amount of the main dopant can be changed at equal intervals, and an equal interval in the range in which the supply amount of additional dopant can be changed. The method may be a method of selecting m levels and calculating conditions of all the combinations n × m and selecting a combination closest to the target, or an efficient search algorithm such as a conjugate gradient method or a quasi-Newton method may be used.

  Thus, by finding the optimum combination of the main dopant and the additional dopant using the resistance distribution prediction formula prepared in the computer, the number of experiments in an actual apparatus can be greatly reduced.

  However, the resistance distribution prediction formula prepared on the computer cannot always accurately predict the resistance distribution. This is because there is a possibility that unexpected parameters may exist in addition to the parameters selected when creating the prediction formula, and the values of the selected parameters and the measured values of the resistance distribution include errors, respectively. This is because it is difficult to create a simple prediction formula.

  In such a case, it is preferable to correct the prediction formula by comparing the prediction based on the resistance distribution prediction formula with the experimental result. That is, a thin film is formed using the semiconductor thin film growth apparatus, and an actual measurement process for measuring the resistance distribution of the formed thin film, and parameter values that are the same conditions as the thin film formation conditions actually measured in the actual measurement process And a correction step of correcting the prediction formula so that an expected resistance distribution obtained by substituting into the measured resistance distribution approximates to the actually measured resistance distribution obtained in the actual measurement step, and in the supply amount determination step, the correction step The optimum supply amount is determined based on the corrected prediction formula that is the corrected prediction formula, and in the thin film forming step, a thin film is formed according to the optimum supply amount determined based on the corrected prediction formula. Also good.

  In the correction step, for example, the term of the gas phase concentration distribution of the dopant added intentionally in the resistance distribution prediction formula is multiplied by the coefficient k1, and the term of the autodope gas phase concentration distribution is multiplied by the coefficient k2, and the experimental results (actual resistance distribution measured) The prediction formula is corrected by determining the coefficients k1 and k2 so as to most closely match. By using this corrected prediction formula (corrected prediction formula), the optimum supply amount of the main dopant and the additional dopant can be determined more accurately, and a thin film having a resistance distribution closer to the target resistance distribution can be obtained. .

In the thin film formation system according to the present invention, the main source gas is divided into a plurality of regions and supplied to the reaction chamber, and the first dopant supply line and the second dopant supply line are connected to different divided regions. A thin film growth apparatus;
A first supply amount that is a supply amount of a dopant supplied from the first dopant supply line to the reaction chamber, and a second supply amount that is a supply amount of a dopant supplied from the second dopant supply line to the reaction chamber. Storage means for storing an expected expression that gives an expected resistance distribution of the thin film, including parameters that affect the resistance distribution of the thin film including the supply amount;
Setting means for setting a value of another parameter that is the parameter other than the first supply amount and the second supply amount and a target resistance distribution of the thin film;
The first supply amount and the second supply amount that minimize the difference between the predicted resistance distribution given by the prediction formula fixed to the set value of the other parameter set by the setting means and the target resistance distribution set by the setting means. A supply amount determining means for determining an optimum supply amount that is a combination of the supply amounts,
The semiconductor thin film growth apparatus is characterized in that a thin film is formed on a main surface of a semiconductor substrate by vapor phase growth in the reaction chamber according to the set value in the other parameters and the optimum supply amount.

  According to the thin film formation system of the present invention, as in the thin film formation method of the present invention, a thin film having a resistance distribution close to the target resistance distribution is obtained compared to the conventional method in which the supply amount of the dopant is determined by trial and error. Can be easily obtained.

1 is a side sectional view of an epitaxial growth apparatus 1. FIG. 1 is a top view of an epitaxial growth apparatus 1. FIG. 3 is a block diagram illustrating a schematic configuration of a computer 4. FIG. It is the flowchart which showed the procedure of thin film formation of this invention. It is the flowchart which showed the preparation procedure of resistance distribution prediction formula. It is the figure which showed the influence of the wafer temperature of the amount of auto dope. It is the figure which showed the influence of the wafer concentration (wafer resistance) of the amount of auto dope. It is the figure which showed the influence of the wafer back surface process of the amount of auto dope. It is the figure which showed the influence of the susceptor shape of the amount of auto dope. It is the figure which illustrated the simulation result of the field of the flow of main source gas. It is the figure which showed distribution of the dopant added intentionally flowing through the AA cross section of FIG. It is the figure which illustrated expected resistance distribution of an epitaxial thin film. It is the figure which illustrated the measured value of resistance distribution of an epitaxial thin film. It is the flowchart which showed the procedure of correction | amendment of resistance distribution prediction type | formula. It is the figure which illustrated measured value of resistance distribution of an epitaxial thin film obtained after amendment of resistance distribution prediction formula. It is the figure which illustrated the experimental result of resistance distribution adjustment of the conventional epitaxial thin film. It is the figure which showed the typical resistance distribution of the epitaxial thin film influenced by the auto dope. It is the figure which showed resistance distribution of the epitaxial thin film at the time of adding an additional dopant to the wafer center part.

  Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a side sectional view of a single wafer epitaxial growth apparatus 1 (vapor phase growth apparatus) as a semiconductor thin film growth apparatus. FIG. 2 is a top view of the epitaxial growth apparatus 1 showing the main part of FIG. Before describing the thin film forming method of the present invention, the configuration of the epitaxial growth apparatus 1 will be described with reference to FIGS.

  As shown in FIG. 1, the epitaxial growth apparatus 1 has an injector 50 (gas supply port) disposed on one end side (left end side in FIG. 1) in the horizontal direction, and is opposite to the side on which the injector 50 is disposed (FIG. 1). In the right end side, the reaction chamber 2 is provided with a gas discharge port 36. The reaction chamber 2 includes a lower case 21 and an upper case 22. The main raw material gas G for forming the epitaxial thin film is supplied from the gas supply line 6 (gas supply pipe) through the injector 50 into the reaction chamber 2 and rotated and held substantially horizontally in the internal space of the reaction chamber 2. After flowing along the main surface of the silicon wafer W as a semiconductor substrate, the gas is discharged from the gas discharge port 36 through the discharge pipe 7. That is, the main raw material gas G flows from the injector 50 toward the gas discharge port 36 in a substantially horizontal direction.

The main source gas G is for vapor-phase growth of a silicon single crystal thin film as an epitaxial thin film on the silicon wafer W, and is a silicon compound such as SiHCl ₃ (TCS, trichlorosilane), SiCl ₄ , SiH ₂ Cl _2, etc. Selected from. In the main raw material gas G, B ₂ H ₆ or PH ₃ as a dopant gas, H ₂ , N ₂ , Ar, or the like as a diluent gas are appropriately blended.

  In the internal space of the reaction chamber 2, a disc-shaped susceptor 12 that is rotationally driven by a motor 13 around a vertical rotation axis O is disposed horizontally, and a silicon epitaxial wafer is placed in a shallow counterbore 12 b formed on the upper surface 12 a thereof. Only one silicon wafer W for manufacturing is disposed. The silicon wafer W is arranged on the susceptor 12 so that the center thereof is at the position of the rotation axis O. The silicon wafer W has a diameter of, for example, 100 mm or more. Further, lamps 11 for heating the silicon wafer W to an epitaxial growth temperature (for example, 900 to 1200 ° C.) at the time of epitaxial growth are arranged at predetermined intervals above and below the reaction chamber 2 corresponding to the arrangement region of the silicon wafer W. For example, a halogen lamp is employed as the lamp 11.

  As shown in FIGS. 1 and 2, a bank member 23 is arranged in the internal space of the reaction chamber 2 so as to surround the susceptor 12. The bank member 23 is arranged in such a positional relationship that the upper surface 23a of the bank member 23 substantially coincides with the upper surface 12a of the susceptor 12 (the main surface of the wafer W). As shown in FIG. 1, the injector 50 is provided so as to face the main surface 23 b of the bank member 23. The main source gas G from the injector 50 hits the main surface 23b and rides on the upper surface 23a side, then flows along the main surface of the wafer W on the susceptor 12, leaving an epitaxial thin film on the wafer W, and discharging gas It is collected in the discharge pipe 7 through the guide member 25 and discharged.

  An imaginary center line perpendicular to the rotational axis O and along the flow direction of the gas supplied from the injector 50 is a horizontal reference line HSL (see FIGS. 1 and 2), which is parallel to the main surface of the wafer W. When the direction orthogonal to the horizontal reference line HSL is the width direction WD (see FIG. 2), the injector 50 is provided with a certain extent in the width direction WD as shown in FIG. The spreading width is approximately the same as the diameter of the susceptor 12. The injector 50 is provided symmetrically in the width direction WD with respect to the horizontal reference line HSL. Further, the injector 50 is divided into a plurality of parts in the width direction WD. Specifically, the injector 50 includes an inner injector 51 provided inside the width direction WD (near the region through which the horizontal reference line HSL passes in the region on the left end side of the reaction chamber 2 in FIG. 2), and the inner injector 51. It consists of two outer injectors 52 provided on the outside. The inner injector 51 is an injector that is provided at a position facing the inner region (the central portion 101 of the wafer W (see FIG. 2)) in the width direction WD of the susceptor 12 and supplies gas toward the inner region. The outer injector 52 is an injector that is provided at a position facing the outer region (the peripheral portion 102 of the wafer W (see FIG. 2)) in the width direction WD of the susceptor 12 and supplies gas toward the outer region.

  A gas guide member 24 in which a gas guide space 24s is formed is provided between the injector 50 and the reaction chamber 2 (the bank member 23). The gas guide member 24 is divided into a plurality of parts in the width direction WD (having a plurality of internal spaces 24 s divided in the width direction WD). The gas from each injector 51, 52 is uniformly distributed in the width direction WD by the divided gas guide member 24 and supplied into the reaction chamber 2.

  The injector 50 is connected to a gas supply line 6 (gas pipe) that guides main raw material gas (silicon source gas, carrier gas) and dopant gas to the injector 50. As shown in FIG. 2, the gas supply line 6 includes a main source gas supply line 61 for supplying main source gas, an inner gas supply line 62 connected to the inner injector 51, and an outer side connected to the outer injector 52. The gas supply line 63 includes a first dopant supply line 64 that supplies a dopant gas, and a second dopant supply line 65 that also supplies the dopant gas. The main source gas supply line 61 is provided so as to branch into an inner gas supply line 62 and an outer gas supply line 63 on the way. That is, the main source gas flowing in the main source gas supply line 61 is supplied into the reaction chamber 2 from both the inner injector 51 and the outer injector 52 via the inner gas supply line 62 and the outer gas supply line 63. It has become.

  The first dopant supply line 64 is connected to the main source gas supply line 61. Therefore, the dopant gas flowing through the first dopant supply line 64 (hereinafter referred to as the main dopant gas) is supplied into the reaction chamber 2 from both the inner injector 51 and the outer injector 52 in the same manner as the main source gas. It has become.

  The second dopant supply line 65 is connected to the inner gas supply line 62. Therefore, a dopant gas (hereinafter referred to as an additional dopant gas) flowing through the second dopant supply line 65 is supplied from the inner injector 51 into the reaction chamber 2. As described above, since the influence of auto-doping is usually large in the peripheral portion 102 of the wafer W, if only the main dopant is supplied, the resistance of the central portion 101 may be larger than the resistance of the peripheral portion 102 ( FIG. 17). The second dopant supply line 65 is provided to supply additional dopant to the central portion 101 of the wafer W via the inner injector 51 and to reduce the resistance of the central portion 101 in order to compensate for the autodoping effect. Supply line.

  The inner gas supply line 62 and the outer gas supply line 63 are provided with flow rate adjusting valves 31 and 32 (for example, needle valves) for adjusting the gas flow rate flowing through the lines 62 and 63, respectively. The flow rate adjusting valves 31 and 32 allow the gas flow rate supplied from the inner injector 51 and the gas flow rate supplied from the outer injector 52 to be the same or different. For example, when the film thickness of the central part 101 is smaller than that of the peripheral part 102, the film thickness of the central part 101 can be reduced by increasing the opening degree of the flow rate adjusting valve 31 provided in the inner gas supply line 62. The film thickness of the portion 102 can be made equal.

  In the present embodiment, the main dopant gas and the additional dopant corresponding to the thin film formation conditions (parameters) are set so that the in-plane resistance distribution of the epitaxial thin film formed on the main surface of the wafer W becomes a distribution close to the target resistance distribution. A system that automatically obtains the optimum supply amount of dopant gas (calculation by a computer) has been constructed. FIG. 3 is a block diagram illustrating a schematic configuration of the computer 4 as the system. 1 and FIG. 2 and the computer 4 of FIG. 3 constitute the “thin film forming system” of the present invention.

  The computer 4 includes a display 42, a storage device 43, a condition input unit 44, and a processing unit 41 connected thereto. The display 42 is a display unit that displays various types of information in accordance with commands from the processing unit 41. Specifically, for example, thin film formation conditions (target resistance distribution, wafer temperature, gas flow rate, etc.) input from the condition input unit 44 are displayed. It displays the measured value of the resistance distribution of the formed thin film, and displays the optimum supply amount of dopant gas and the expected resistance distribution calculated by the processing unit 41.

  The storage device 43 is a hard disk, flash memory, or the like that stores various types of information. The storage device 43 stores a resistance distribution prediction formula described later. The storage device 43 corresponds to the “storage means” of the present invention. The condition input unit 44 includes various conditions (parameters other than the supply amount of the dopant gas, which are necessary for predicting the resistance distribution based on the resistance distribution prediction formula stored in the storage device 43. For example, the target resistance distribution, wafer temperature, silicon (Source gas flow rate, carrier gas flow rate, etc.) are input to the computer 4. The condition input unit 44 is an input unit that is input by an operator such as a keyboard or a mouse, or is currently set in the epitaxial growth apparatus 1 by communication from the epitaxial growth apparatus 1 (or a control device that controls the operation of the epitaxial growth apparatus 1). It may be a communication part that acquires the condition that

  The processing unit 41 predicts the resistance distribution of the epitaxial thin film formed by the epitaxial growth apparatus 1 based on the conditions (parameters) input from the condition input unit 44 and the resistance distribution prediction formula stored in the storage device 43, A process for calculating the optimum gas supply amount (hereinafter referred to as resistance adjustment support process) is executed. The processing unit 41 includes a CPU 411, a ROM 412, a RAM 413, and the like, and the resistance adjustment support process is realized by the CPU 411 executing a program (resistance adjustment software) stored in the ROM 412.

  In order for the processing unit 41 to execute the resistance adjustment support process, it is necessary to prepare a resistance distribution prediction formula to be stored in the storage device 43 in advance. Below, the procedure of thin film formation in this embodiment including preparation of the resistance distribution prediction formula and resistance adjustment support processing executed by the processing unit 41 will be described. FIG. 4 is a flowchart showing the procedure.

  First, a resistance distribution prediction formula for predicting the in-plane resistance distribution of the epitaxial thin film formed by the epitaxial growth apparatus 1 is created (S1). The in-plane resistance distribution of the epitaxial thin film is determined by the in-plane distribution of the dopant concentration of the epitaxial thin film. That is, a portion with a low dopant concentration shows a large resistance value, and a portion with a high dopant concentration shows a small resistance value. The dopant concentration of the epitaxial thin film includes the air concentration distribution Cint (r) of the dopant added intentionally (dopant supplied from the injector 50), the air concentration distribution Cauto (r) of the dopant by auto-doping, and the epitaxial concentration of the dopant. It is determined by the incorporation rate Z into the thin film and the growth rate GR of the epitaxial thin film. Specifically, the intentionally added air concentration distribution Cint (r) of the dopant and the air concentration distribution Cauto (r) of the dopant by auto-doping are multiplied by the intake rate Z and divided by the growth rate GR. Becomes the dopant concentration distribution (resistance distribution) of the epitaxial thin film. That is, the dopant concentration distribution (resistance distribution) of the epitaxial thin film = (Cint (r) + Cato (r)) × Z / GR. Therefore, in S1, a formula for predicting the resistance distribution (r) is created from these parameters Cint (r), Cato (r), Z, GR. In addition, the process of S1 corresponds to the “forecast formula creation process” of the present invention.

  Specifically, for example, a resistance distribution prediction formula is created according to the procedure of FIG. That is, first, a function f1 (first function) for obtaining the growth rate GR from a parameter (first parameter) that affects the growth rate GR of the epitaxial thin film is created (S11). The step S11 corresponds to the “first creation step” of the present invention. Specifically, as a method of creating the function f1, first, it is examined how the growth rate GR changes when the first parameter is changed. In this embodiment, the wafer temperature T, the carrier hydrogen (carrier gas) flow rate Fc, and the silicon source gas (for example, TCS) flow rate Fs are selected as the first parameters. Each parameter T, Fc, Fs is divided into several levels, and for each level, an epitaxial thin film is actually formed using the epitaxial growth apparatus 1, or virtually by computer simulation simulating the epitaxial growth apparatus 1. An epitaxial thin film is formed. Then, the growth rate of each formed epitaxial thin film is calculated. Table 1 below illustrates the experimental results.

  In the example of Table 1, the wafer temperature T is divided into two levels, the carrier hydrogen flow rate Fc is divided into three levels, the silicon source gas flow rate Fs is divided into two levels, and the total 12 patterns of experimental results are shown. Yes. In Table 1, the nine-point average film thickness (average film thickness at nine predetermined points of the formed epitaxial thin film) is shown as a value correlated with the growth rate. The growth rate is calculated by dividing the nine-point average film thickness by the growth time.

In S11, after performing the experiment as shown in Table 1, a function f1 (T, Fc, Fs) for obtaining the growth rate GR from an arbitrary value of each parameter T, Fc, Fs is created based on the experiment result. . Any method may be used to obtain the function f1 from the experimental results. For example, the theoretical formula of the growth rate GR using the parameters T, Fc, and Fs as variables is analytically derived, and most agrees with the experimental results in Table 1. There is a method of correcting the theoretical formula so that the coefficient part of the theoretical formula is set. For example, if the growth rate GR is analytically determined by a model in which the growth rate GR is determined by only the diffusion phenomenon and the growth rate GR is constant regardless of the distance from the wafer and continues to an infinite height, GR = k C · (D · u / x) ^0.5 (GR: growth rate, k: coefficient, C: source gas concentration, D: diffusion coefficient, u: gas velocity, x: position) can be derived. Since the source gas concentration C correlates with the gas flow rates Fc and Fs and the diffusion coefficient D correlates with the wafer temperature T, the equation is also an equation with the parameters T, Fc and Fs as variables. Since the experimental results in Table 1 do not fit the derived equations, the above equation is replaced with GR = k · C ^p · (D · u / x) ^q , and k, p so as to best match the experimental results. , Q are obtained, and the finally obtained expression is defined as a function f1 (T, Fc, Fs).

  Next, a function f2 (second function) for obtaining the air concentration distribution Cato (r) of the dopant by autodoping from the parameter (second parameter) affecting the autodoping amount is created (S12). The step S12 corresponds to the “second creation step” of the present invention. Specifically, as a method of creating the function f2, first, it is examined how the auto-doping amount changes when the second parameter is changed. In this embodiment, the wafer temperature T, the carrier hydrogen flow rate Fc, the silicon source gas flow rate Fs, the susceptor shape S, the wafer resistance Wr, and the wafer back surface treatment Wb are selected as the second parameters. The wafer temperature T, the susceptor shape S, the wafer resistance Wr, and the wafer back surface treatment Wb are parameters that affect the amount of dopant released from the side surface and the back surface of the wafer. The carrier hydrogen flow rate Fc and the silicon source gas flow rate Fs are parameters that affect the amount taken into the epitaxial thin film out of the amount of dopant released from the wafer.

  Then, each parameter T, Fc, Fs, S, Wr, Wb is divided into several levels, and the auto-doping amount is measured or simulated for each level. 6 to 9 illustrate the experimental results (actually measured values). Specifically, FIG. 6 to FIG. 9 show that an epitaxial thin film is formed using the epitaxial growth apparatus 1, the dopant concentration of the formed epitaxial thin film is used as the autodoping amount, and the in-plane distribution of the dopant concentration (width from the wafer center). The dopant concentration with respect to the distance in the direction WD). In the experiments of FIGS. 6 to 9, the dopant gas is not supplied from the injector 50. Moreover, regarding the measuring method of dopant concentration, the resistance of the formed epitaxial thin film was measured by CV method, SPV method, etc., and the dopant concentration (auto-doping amount) of the epitaxial thin film was calculated from the measured resistance.

  FIG. 6 is a diagram showing the influence of the wafer temperature T on the autodoping amount. In FIG. 6, the wafer temperature T is divided into two levels. In the experiment of FIG. 6, the values of the parameters Fc, Fs, S, Wr, and Wb other than the wafer temperature T are fixed.

  FIG. 7 is a diagram showing the influence of the dopant concentration (hereinafter referred to as wafer concentration) originally contained in the silicon wafer W to be used as a value correlated with the wafer resistance Wr. In FIG. 7, the wafer concentration is divided into two levels. Specifically, the result of level 1 shows the dopant concentration of the epitaxial thin film when the epitaxial thin film is formed using the silicon wafer having the wafer concentration divided into P +. The result of level 2 shows the dopant concentration of the epitaxial thin film when the epitaxial thin film is formed using the silicon wafer in which the wafer concentration is divided into P ++. In the experiment of FIG. 7, the values of parameters T, Fc, Fs, S, and Wb other than the wafer concentration (wafer resistance Wr) were fixed. As shown in FIG. 7, when the P ++ wafer is used, the dopant concentration (auto-doping amount) increases more than when the P ++ wafer is used.

  FIG. 8 is a diagram showing the influence of the wafer back surface processing Wb. In FIG. 8, the back surface treatment Wb is divided into two levels. Specifically, the level 1 result indicates the dopant concentration of the epitaxial thin film when the epitaxial thin film is formed using the silicon wafer with the back surface treatment. These results show the dopant concentration of the epitaxial thin film when the epitaxial thin film is formed using the silicon wafer without the back surface treatment. In Level 1, specifically, the SiO2 film is formed on the backside of the wafer by the CVD method as the backside treatment. However, the backside treatment other than the SiO2 film is performed, or the thickness of the formed SiO2 film is varied. Thus, the influence of the back surface processing Wb may be examined. In the experiment of FIG. 8, the values of parameters T, Fc, Fs, S, and Wr other than the back surface processing Wb are fixed. As shown in FIG. 8, it can be seen that the amount of autodoping can be suppressed by using a wafer with a backside treatment.

  FIG. 9 shows the influence of the susceptor shape S. FIG. In FIG. 9, the susceptor shape S is divided into two levels. Specifically, the result of level 1 is that when an epitaxial thin film is formed using a susceptor 12 of a type in which the bottom surface of the counterbore 12b (see FIG. 1) is closed. The dopant concentration of the epitaxial thin film is shown, and the result of level 2 shows the dopant concentration of the epitaxial thin film when the epitaxial thin film is formed using the susceptor 12 of the type in which holes are formed in the bottom surface of the counterbore 12b. As disclosed in Japanese Patent Application Laid-Open No. 2003-289044, when a hole is formed in the bottom surface of the susceptor, part of the source gas escapes from the hole to the back surface side of the susceptor, and the dopant released from the wafer side surface or back surface is The amount of auto-doping can be suppressed because it is released to the back side of the susceptor by riding on the gas flow that escapes to the back side. In FIG. 9, the dopant concentration (auto-doping amount) is lower in level 2 (susceptor in which holes are formed) than in level 1. In the experiment of FIG. 9, the values of the parameters T, Fc, Fs, Wr, and Wb other than the susceptor shape S are fixed.

  In any of the experimental results shown in FIGS. 6 to 9, the dopant concentration (auto-doping amount) tends to increase as the distance from the wafer center increases. That is, the experimental results of FIGS. 6 to 9 indicate that the influence of auto-doping is increased in the peripheral portion 102 (see FIG. 2) of the wafer W.

  Note that the carrier hydrogen flow rate Fc and the silicon source gas flow rate Fs as the second parameters are parameters that affect the growth rate GR, and the autodoping amount also changes according to the growth rate GR. Was used as an experimental result of the influence of the carrier hydrogen flow rate Fc and the silicon source gas flow rate Fs on the autodope amount.

  In S12, after performing the experimental results shown in FIGS. 6 to 9 and Table 1, based on the experimental results, the value of the dopant by auto-doping is determined from an arbitrary value of each parameter T, Fc, Fs, S, Wr, and Wb. A function f2 (r, T, Fc, Fs, S, Wr, Wb) for obtaining the medium concentration distribution Cato (r) is created. R is a parameter indicating the position in the gas phase. Any method may be used to obtain the function f2 from the experimental results. For example, similarly to S11, the theoretical formula of the air concentration distribution Cato (r) using the parameters T, Fc, Fs, S, Wr, and Wb as variables. Is derived analytically, and the theoretical formula is corrected so as to best match the experimental result (the coefficient portion of the theoretical formula is set), thereby obtaining the function f2.

  Next, in order to obtain the air concentration distribution of the dopant added intentionally, parameters (first parameter) affecting the region where the main source gas (silicon source gas, carrier gas) dividedly supplied by the injector 50 flows in the reaction chamber 2. A function f3 (third function) for obtaining the flow region from the three parameters) is created (S13). The step S13 corresponds to the “third creation step” of the present invention. Specifically, as a method of creating the function f3, first, it is examined how the flow region changes when the third parameter is changed. In the present embodiment, as a third parameter, a flow rate adjustment valve 31 for distributing the carrier hydrogen flow rate Fc and the main source gas composed of a mixed gas of carrier hydrogen and silicon source gas to the inner region and the outer region of the reaction chamber 2, The dial values Vin and Vout of 32 and the rotation speed Rot of the susceptor 12 were selected.

  The dial value Vin indicates the opening degree of the flow rate adjustment valve 31 provided in the inner gas supply line 62. As the dial value Vin increases, the opening degree of the flow rate adjustment valve 31 increases, and the flow rate flowing through the inner gas supply line 62 and the flow rate of the gas supplied from the inner injector 51 increase. Similarly, the dial value Vout indicates the opening degree of the flow rate adjustment valve 32 provided in the outer gas supply line 63. As the dial value Vout increases, the opening degree of the flow rate adjustment valve 32 increases, and the flow rate flowing through the outer gas supply line 63 and the flow rate of the gas supplied from the outer injector 52 increase. The reason why the rotation speed Rot of the susceptor 12 is included in the third parameter is that the gas in the reaction chamber 2 is dragged by the rotation of the susceptor 12, and the amount of gas dragged changes according to the rotation speed Rot.

  Then, each parameter Fc, Vin, Vout, and Rot is divided into several levels, and the fluid simulation of the flow region of the main raw material gas is performed for each level. The fluid simulation can be performed by commercially available fluid analysis software. For example, ANSYS Fluent can be used as the fluid analysis software. When performing the fluid simulation, the structure of the epitaxial growth apparatus 1 (the injector 50 is divided into the inner injector 51 and the outer injector 52, the shape and size of the reaction chamber 2, etc.) is subjected to the fluid simulation. This is input to a computer to simulate the structure of the epitaxial growth apparatus 1.

  FIG. 10 is a diagram illustrating a simulation result of a flow region at a certain value of each parameter Fc, Vin, Vout, and Rot. In FIG. 10, the flow of the main raw material gas is shown in gradation so as to overlap with the top view of the epitaxial growth apparatus 1 (top view of the reaction chamber 2). In FIG. 10, the flow region of the main source gas supplied from the inner injector 51 (hereinafter referred to as the inner gas) is shown in dark gray (color close to black), and the main source gas supplied from the outer injector 52 (hereinafter referred to as the “inner gas”). The flow region of the outer gas) is shown in light gray, and the flow region of the mixed gas of the inner gas and the outer gas is shown in gray close to white. FIG. 10 shows a state in which the inner gas and the outer gas are gradually mixed as they flow to the discharge side (the gray area close to white increases). Further, FIG. 10 shows an example in which the susceptor 12 is rotating clockwise, and shows a state in which the flow of the main raw material gas is dragged (not flowing straight) by the rotation.

  In the present embodiment, the influence distribution Fin (r) of the inner gas is obtained as an index indicating the flow region of the main raw material gas. For this purpose, first, the simulation result of FIG. 10 is digitized. Specifically, when the inner gas region (the region where the outer gas is not mixed) is represented by 1 and the outer gas region (the region where the inner gas is not mixed) is represented by 0, the mixed gas region is represented by the inner gas. It can be expressed by a numerical value between 1 and 0 depending on how mixed. Specifically, for example, a region where the inner gas is mixed at 80% and the outer gas is mixed at a rate of 20% can be represented by 0.8.

  Further, since the wafer W (susceptor 12) is rotating, the region whose distance from the center of the wafer W is r is an amount obtained by averaging the above numerical simulation results with a circle having a radius r, specifically a radius. It is affected by the inner gas by the amount obtained by line integration with r and divided by the circumference of the circle. The average amount distribution is defined as an influence distribution Fin (r) of the inner gas. As described above, in S13, the simulation of FIG. 10 is performed for each level of the third parameters Fc, Vin, Vout, and Rot, and the influence distribution Fin (r) is calculated. By interpolating the distribution Fin (r), a function f3 (r, Fc, Vin, Vout, Rot) for obtaining the influence distribution Fin (r) from an arbitrary value of each parameter Fc, Vin, Vout, Rot is created.

The influence distribution Fin (r) = function f3 (r, Fc, Vin, Vout, Rot) created in S13 indicates how the dopant supplied from the injector 50 flows in the reaction chamber 2. Therefore, by adding information of the main dopant supply amount Dm and the additional dopant supply amount Da to the influence distribution Fin (r), the intentional addition of the dopant (supplied from the injector 50) A function f4 (Fin (r), Dm, Da) for obtaining the density distribution Cint (r) is created (S14). Specifically, assuming that the units of the supply amounts Dm and Da are the dopant concentration (mol / m ³ ), the function f4 of the air concentration distribution Cint (r) is f4 = Fin (r) · Dm + (1−Fin (R)) · Da. FIG. 11 illustrates the distribution Cint (r) of the intentionally added dopant flowing through the AA cross section of FIG. 10 at a certain value of the third parameter Fc, Vin, Vout, Rot and the supply amounts Dm, Da. . The horizontal axis in FIG. 11 indicates the distance r from the center of the wafer W, and the vertical axis indicates the dopant concentration (value normalized by the maximum value of the dopant concentration (relative value)). 10 is a line extending in the width direction WD passing through the center of the wafer W. In the example of FIG. 11, the supply amount Dm of the main dopant (the amount of dopant from the first dopant supply line 64 (see FIG. 10)) is zero, and the distribution of only the additional dopant is shown. As shown in FIG. 11, it can be seen that a large amount of additional dopant is supplied to the region near the center of the wafer W. The step S14 corresponds to the “fourth creation step” in the present invention.

  Next, a dopant incorporation rate Z into the epitaxial thin film is calculated (S15). Specifically, as described above, the dopant concentration distribution (resistance distribution) of the epitaxial thin film is expressed by the following formula: dopant concentration distribution (resistance distribution) = (Cint (r) + Cato (r)) × Z / GR, In S15, the capture speed Z is experimentally calculated from the equation. The left side of the above equation (dopant concentration distribution of the epitaxial thin film) can be obtained by actually forming the epitaxial thin film using the epitaxial growth apparatus 1 and measuring the resistance of the formed epitaxial thin film. The air concentration distributions Cint (r) and Cato (r) in the above formula are obtained in steps S12 to S14. The growth rate GR in the above formula is obtained in step S11. Therefore, since the unknown is only the capture rate Z, the capture rate Z can be obtained by substituting the known number (dopant concentration distribution of the epitaxial thin film, Cint (r), Cato (r), GR) into the above formula. If the theoretical value of the capture speed Z is known, the theoretical value may be used.

  Note that only a small portion of the dopant injected into the reaction chamber 2 is taken into the epitaxial thin film. That is, in the dopant incorporation into the epitaxial thin film, the dopant transport process is not rate-limiting, and the incorporation process is rate-limiting. Therefore, the uptake speed Z is a function of only the temperature T without depending on the flow rate or concentration. The temperature dependence of Z can be obtained by performing the experiment of S15 at several levels with different temperatures T.

  In addition, the order which implements the process of S11-S15 is not limited to the order of FIG. That is, the function f1 of the growth rate GR, the function f2 of the air concentration distribution Cauto (r) of the dopant by autodoping, the function f4 of the air concentration distribution Cint (r) of the dopant added intentionally, and the uptake speed Z You may ask in order.

  Next, based on the functions f1, f2, and f4 obtained in S11 to S15, and the incorporation rate Z, a resistance distribution prediction formula Rcalc (r) = f5 (GR, Cint () r), Cato (r), Z) are created (S16). Specifically, an equation represented by Rcalc (r) = (Cint (r) + Cato (r)) × Z / GR is used as a resistance distribution prediction equation. In the present embodiment, the wafer temperature T is used as the capture speed Z, and Rcalc (r) = f5 (GR, Cint (r), Cato (r), T) is obtained in S16. Since the resistance distribution prediction equation parameters GR, Cint (r), and Cato (r) are created from directly controllable primary parameters T, Fc, Fs, S, Wr, Wb, Vin, Vout, Dm, and Da. The resistance distribution prediction formula is an expression for obtaining the predicted resistance distribution Rcalc (r) from the primary parameters T, Fc, Fs, S, Wr, Wb, Vin, Vout, Dm, Da. The steps S15 and S16 correspond to the “fifth creation step” of the present invention.

  Returning to the description of FIG. 4, after the resistance distribution prediction formula is created in S1, the obtained resistance distribution prediction formula is stored in the storage device 43 (see FIG. 3). Then, when actually forming an epitaxial thin film using the epitaxial growth apparatus 1, the process of S2-S4 is implemented. That is, the operator uses the condition input unit 44 to set parameters T, Fc, Fs, S, Wr, Wb, Vin, Vout (hereinafter referred to as other parameters) which are thin film formation conditions other than the dopant supply amounts Dm, Da. ) And the target resistance distribution are input (S2). Then, the processing unit 41 sets (stores) the other parameters and the target resistance distribution input from the condition input unit 44 as thin film formation conditions in the RAM 413 or the like (S2). In the following description, the thin film formation conditions shown in Table 2 below are set in S2. The condition input unit 44 and the processing unit 41 that executes the processing of S2 correspond to the “setting unit” of the present invention.

  Next, based on the thin film formation conditions set in S2 (Table 2) and the resistance distribution prediction formula stored in the storage device 43, the processing unit 41 optimizes the dopant supply amounts Dm and Da (optimum supply amount). Is calculated (S3). Specifically, the set values of other parameters T, Fc, Fs, S, Wr, Wb, Vin, and Vout (values in Table 2) are substituted into the resistance distribution prediction formula. Then, a search is made for a combination of supply amounts Dm and Da that minimizes the difference in distance between the predicted resistance distribution given by the resistance distribution prediction formula after substitution and the target resistance distribution set in S2. As a method for defining the distance between the predicted resistance distribution and the target resistance distribution, for example, several points are selected on the wafer surface, and the square sum of the difference between the predicted resistance and the target resistance at each point is calculated as the distance. As a method for searching for an optimal combination of the supply amounts Dm and Da, for example, a changeable range of the main dopant supply amount Dm is set at n levels at equal intervals, and a changeable range of the supply amount Da of the additional dopant is set. It is possible to select m levels at equal intervals, calculate the conditions of all combinations n × m, and select the combination closest to the target, or use an efficient search algorithm such as conjugate gradient method or quasi-Newton method. Also good. In this embodiment, the quasi-Newton method is employed, and Dm = 252.7 sccm and Da = 58.0 sccm are obtained as the optimum values of the supply amounts Dm and Da. The step S3 corresponds to the “supply amount determining step” of the present invention. The processing unit 41 that executes the process of S3 corresponds to the “supply amount determining means” of the present invention.

  FIG. 12 shows an expected resistance distribution obtained by substituting this optimum value and the set value of Table 2 into the resistance distribution prediction formula. The horizontal axis in FIG. 12 indicates the distance from the wafer center in the width direction WD, and the vertical axis indicates the resistivity. FIG. 12 also shows the target resistance distribution (value in Table 2) set in S2. As shown in FIG. 12, the expected resistance distribution is very close to the target resistance distribution.

  Next, the worker actually vapor-phases an epitaxial thin film on the main surface of the wafer using the epitaxial growth apparatus 1 under the conditions of the set values set in S2 and the optimum values of the supply amounts Dm and Da obtained in S3. (S4). The process of S4 corresponds to the “thin film forming process” of the present invention. FIG. 13 shows measured values of the resistance distribution of the epitaxial thin film obtained in S4. The horizontal and vertical axes in FIG. 13 are the same as those in FIG. The actually measured resistance distribution in FIG. 13 is slightly different from the target resistance distribution as compared with the predicted resistance distribution in FIG. 12, but the resistance is very close to the target resistance distribution for the first epitaxial thin film formation. A distributed epitaxial thin film can be easily obtained without trial and error. The above is the thin film formation method of this embodiment. When the resistance distribution prediction formula is once created in the step S1, the steps S2 to S4 may be performed from the next time.

  On the other hand, in order to make the resistance distribution of the epitaxial thin film obtained in step S4 closer to the target resistance distribution, the step shown in FIG. 14 may be performed after step S4 to correct the resistance distribution prediction formula. Specifically, the resistance distribution of the epitaxial thin film formed in S4 is measured (S5, FIG. 13). The step S5 corresponds to the “actual measurement step” of the present invention. Next, the expected resistance distribution obtained by substituting the parameter values (that is, the set values in Table 2 and the optimum values obtained in S3) that are the same as the epitaxial thin film formation conditions measured in S5 into the resistance distribution prediction formula However, the resistance distribution prediction formula is corrected so as to approach the resistance distribution actually measured in S5 (S6). Specifically, for example, in the term of the air concentration distribution Cint (r) of the dopant added intentionally in the resistance distribution prediction formula Rcalc (r) = f5 (GR, Cint (r), Cato (r), T) The coefficient k1 is multiplied, and the term of the air concentration distribution Cauto (r) of the dopant by auto-doping is multiplied by the coefficient k2. Then, the coefficients k1 and k2 are determined so that the expected resistance distribution most closely matches the actually measured resistance distribution. As a method of determining the coefficients k1 and k2, for example, a search algorithm such as a conjugate gradient method or a quasi-Newton method may be used. The step S6 corresponds to the “correction step” of the present invention.

  Then, the resistance distribution prediction formula stored in the storage device 43 is updated to the resistance distribution prediction formula corrected in S6 (hereinafter referred to as a correction prediction formula), and supply is performed using the corrected prediction formula in the subsequent thin film formation. The optimum values of the amounts Dm and Da are calculated (S3). Assuming that the set values set in the next S2 are the values in Table 2, in the next S3, Dm = 257.4 sccm and Da = 65.7 sccm were obtained as the optimum values of the supply amounts Dm and Da. An epitaxial thin film is formed according to the optimum value (S4). FIG. 15 shows measured values of the resistance distribution of the epitaxial thin film obtained in step S4. The horizontal and vertical axes in FIG. 15 are the same as those in FIG. The actually measured resistance distribution in FIG. 15 is closer to the target resistance distribution than the first resistance distribution (FIG. 13).

(Examples and comparative examples)
FIG. 16 is a diagram illustrating an experimental result of resistance distribution adjustment of a conventional epitaxial thin film. Specifically, a target resistance distribution is set for each day (experiment day) shown in FIG. 16A, and the supply amount of main dopant and additional dopant is adjusted by trial and error until the target resistance distribution is approached. An experiment was conducted. FIG. 16A shows the transition of the supply amount of the main dopant and the additional dopant. The number of plot points in FIG. 16A indicates the number of times that the supply amount of the main dopant and the additional dopant is adjusted. FIG. 16B shows the transition of the target resistance (resistivity) and the measured resistance distribution (maximum value and minimum value of the resistance distribution). It should be noted that experiments were conducted on dates other than those shown in FIG. 16, and FIG. 16 shows only experimental results on some of the dates.

  For example, the target resistance (= 10 Ω · cm) of the experiment date is 3/15 has changed from the target resistance (= 16 Ω · cm) of 3/14, and the number of adjustments of the dopant in 9/15 is 9 times. ing. As described above, in the conventional method, the number of adjustments of the dopant supply amount increases particularly when the target resistance changes. The average value of the number of adjustments when the target resistance is changed (the average value including the experimental results other than the experimental date in FIG. 16) is 7.0 times (that is, seven epitaxial wafers are obtained before the target resistance is obtained). Need to be manufactured once). When the target resistance does not change (for example, the experiment date of 5/11, 5/13, 5/23, 6/24, and 7/20 in FIG. 16), the previous adjustment result can be used. Although the number of adjustments can be reduced compared to when the resistance changes, the average number of adjustments is still 2.8. On the other hand, in the present invention, the average value of the number of adjustments when the target resistance is changed is 2.0. Therefore, by applying the present invention, the number of adjustments of the dopant supply amount can be greatly reduced as compared with the conventional case.

  As described above, in this embodiment, after the resistance distribution prediction formula is once created, the optimum dopant supply amount corresponding to the set value of the thin film formation conditions (other parameters) and the target resistance distribution is automatically (by the computer). ), An epitaxial thin film having a resistance distribution close to the target resistance distribution can be easily obtained.

  In addition, this invention is not limited to the said embodiment, A various change is possible to the limit which does not deviate from description of a claim. For example, in the above embodiment, the example in which the present invention is applied to a system in which a silicon thin film is vapor-grown on the main surface of a silicon wafer has been described. However, the present invention is also applied to a system for manufacturing an epitaxial wafer other than a silicon epitaxial wafer. it can. In the above-described embodiment, the parameters (other parameters) included in the resistance distribution prediction formula include the wafer temperature T, the carrier hydrogen flow rate Fc, the silicon source gas flow rate Fs, the susceptor shape S, the wafer resistance Wr, the wafer back surface processing Wb, and the valve. Although the dial values Vin and Vout are listed, a resistance distribution prediction formula may be created in consideration of other parameters in addition to these parameters. The method for creating the resistance distribution prediction formula is not limited to the above embodiment, and any method may be used to create the resistance distribution prediction formula.

DESCRIPTION OF SYMBOLS 1 Epitaxial growth apparatus 2 Reaction chamber 12 Susceptor 31, 32 Flow control valve 50 Injector 51 Inner injector 52 Outer injector 61 Main raw material gas supply line 62 Inner gas supply line 63 Outer gas supply line 64 First dopant supply line 65 Second dopant Supply line

Claims (6)

A semiconductor substrate using a semiconductor thin film growth apparatus in which a main source gas is divided into a plurality of regions and supplied to a reaction chamber, and the first dopant supply line and the second dopant supply line are connected to different divided regions A thin film forming method for forming a thin film on the main surface of
A first supply amount that is a supply amount of a dopant supplied from the first dopant supply line to the reaction chamber, and a second supply amount that is a supply amount of a dopant supplied from the second dopant supply line to the reaction chamber. A prediction formula creation process for creating a prediction formula that gives a predicted resistance distribution of the thin film, including parameters that affect the resistance distribution of the thin film including the supply amount;
A difference between an expected resistance distribution given by the prediction formula in which other parameters that are parameters other than the first supply amount and the second supply amount are fixed to a preset setting value, and a preset target resistance distribution of the thin film is A supply amount determining step for determining an optimum supply amount that is a combination of the first supply amount and the second supply amount that are the smallest;
A thin film forming step of forming a thin film on the main surface of the semiconductor substrate by vapor phase growth in the reaction chamber according to the set value in the other parameter and the optimum supply amount;
A thin film forming method comprising:   In the prediction formula creation step, the growth rate of the thin film, the air concentration distribution of the dopant intentionally including the first supply amount and the second supply amount as parameters, and the air concentration distribution of the dopant by autodoping The method for forming a thin film according to claim 1, wherein the prediction equation is created using the parameters as parameters. The prediction formula calculation step includes
Changing the first parameter that affects the growth rate of the thin film, examining the growth rate, and based on the result, creating a first function for calculating the growth rate from the first parameter;
The second parameter that affects the amount of autodoping incorporated into the thin film is changed to investigate the amount of autodoping, and based on the result, a second function is calculated that calculates the air concentration distribution of the dopant by autodoping from the second parameter. A second creation step to
A third function for investigating the flowing region by changing a third parameter affecting the region in which the main source gas supplied in a divided manner flows in the reaction chamber, and calculating the flowing region from the third parameter based on the result A third creation process to create
A fourth creation step of creating a fourth function for calculating an air concentration distribution of the dopant added intentionally by adding the parameters of the first supply amount and the second supply amount to the third function;
The thin film forming method according to claim 1, further comprising a fifth creation step of creating the prediction formula based on the first function, the second function, and the fourth function.   In the fifth creation step, the prediction formula is created based on the incorporation rate of the dopant incorporated into the thin film in addition to the first function, the second function, and the fourth function. Item 4. The method for forming a thin film according to Item 3. An actual measurement step of performing thin film formation using the semiconductor thin film growth apparatus and measuring the resistance distribution of the formed thin film;
The predicted resistance distribution obtained by substituting the value of the parameter that is the same as the thin film formation condition measured in the actual measurement step into the prediction equation approaches the predicted resistance distribution obtained in the actual measurement step. A correction step of correcting the formula,
In the supply amount determination step, the optimum supply amount is determined based on a corrected prediction formula that is the prediction formula corrected in the correction step,
5. The thin film forming method according to claim 1, wherein in the thin film forming step, a thin film is formed according to the optimum supply amount determined based on the correction prediction formula. A semiconductor thin film growth apparatus in which the main source gas is divided into a plurality of regions and supplied to the reaction chamber, and the first dopant supply line and the second dopant supply line are connected to different divided regions;
A first supply amount that is a supply amount of a dopant supplied from the first dopant supply line to the reaction chamber, and a second supply amount that is a supply amount of a dopant supplied from the second dopant supply line to the reaction chamber. Storage means for storing an expected expression that gives an expected resistance distribution of the thin film, including parameters that affect the resistance distribution of the thin film including the supply amount;
Setting means for setting a value of another parameter that is the parameter other than the first supply amount and the second supply amount and a target resistance distribution of the thin film;
The first supply amount and the second supply amount that minimize the difference between the predicted resistance distribution given by the prediction formula fixed to the set value of the other parameter set by the setting means and the target resistance distribution set by the setting means. A supply amount determining means for determining an optimum supply amount that is a combination of the supply amounts,
The semiconductor thin film growth apparatus forms a thin film on a main surface of a semiconductor substrate by vapor phase growth in the reaction chamber according to the set value in the other parameter and the optimum supply amount.