Method of predicting thermal donor formation behavior in silicon wafer, method of evaluating silicon wafer, and method of producing silicon wafer

Abstract

Provided is a method of accurately predicting the thermal donor formation behavior in a silicon wafer, a method of evaluating a silicon wafer using the prediction method, and a method of producing a silicon wafer using the evaluation method. The method of predicting the formation behavior of thermal donors, includes: a first step of setting an initial oxygen concentration condition before performing heat treatment on the silicon wafer for reaction rate equations based on both a bond-dissociation model of oxygen clusters associated with the diffusion of interstitial oxygen and a bonding model of oxygen clusters associated with the diffusion of oxygen dimers; a second step of calculating the formation rate of oxygen clusters formed through the heat treatment using the reaction rate equations; and a third step of calculating the formation rate of thermal donors formed through the heat treatment based on the formation rate of the oxygen clusters.

Claims

1. A method of predicting a formation behavior of thermal donors formed due to oxygen generated when heat treatment is performed on a silicon wafer, comprising: a first step of setting an initial oxygen concentration condition before performing the heat treatment on the silicon wafer for reaction rate equations based on both a bond-dissociation model of oxygen clusters associated with diffusion of interstitial oxygen and a bonding model of oxygen clusters associated with diffusion of oxygen dimers; a second step of calculating a formation rate of oxygen clusters formed through the heat treatment using the reaction rate equations; and a third step of calculating a formation rate of thermal donors formed through the heat treatment based on the formation rate of the oxygen clusters, wherein the reaction rate equations are the following formulae (1) to (4): $\begin{array}{cc}\frac{d\left[O_1\right]}{\mathrm{dt}}=-2{k_{f1}\left[O_1\right]}^2-k_{f1}\left[O_1\right]\left(\underset{n=2}{\overset{M-1}{.Math.}}\left[O_n\right]\right)+2k_{b2}\left[O_2\right]-k_{f2}\left[O_2\right]\left[O_1\right]+\underset{n=3}{\overset{M-1}{.Math.}}{k}_{\mathrm{bn}}\left[O_n\right],& \left(1\right)\\ \frac{d\left[O_2\right]}{\mathrm{dt}}={k_{f1}\left[O_1\right]}^2-k_{f1}\left[O_2\right]\left[O_1\right]-k_{b2}\left[O_2\right]+k_{b3}\left[O_3\right]-k_{f2}\left[O_2\right]\left(\left[O_1\right]+2\left[O_2\right]+\underset{n=3}{\overset{M-2}{.Math.}}\left[O_n\right]\right),& \left(2\right)\\ \frac{d\left[O_n\right]}{\mathrm{dt}}=k_{f1}\left[O_1\right]\left[O_{n-1}\right]-k_{f1}\left[O_1\right]\left[O_n\right]+k_{f2}\left[O_2\right]\left(\left[O_{n-2}\right]-\left[O_n\right]\right)-k_{\mathrm{bn}}\left[O_n\right]+k_{\mathrm{bn}+1}\left[O_{n+1}\right],& \left(3\right)\\ \frac{d\left[O_M\right]}{\mathrm{dt}}=k_{f1}\left[O_1\right]\left[O_{M-1}\right]-k_{f1}\left[O_1\right]\left[O_M\right]+k_{f2}\left[O_2\right]\left[O_{M-2}\right]-k_{\mathrm{bM}}\left[O_M\right],& \left(4\right)\end{array}$ where t represents a time; M is a maximum number of oxygen clusters that become donors; [O.sub.1] and [O.sub.n] represent concentrations of interstitial oxygen O.sub.i and oxygen cluster O.sub.n, respectively; k.sub.f1 is a combination rate coefficient of interstitial oxygen bound to either other interstitial oxygen O.sub.i or oxygen clusters, depending on a heat treatment temperature; k.sub.f2 is a combination rate coefficient of oxygen dimers bound to either other interstitial oxygen O.sub.i or oxygen clusters, depending on the heat treatment temperature; k.sub.b2 is a dissociation rate constant of interstitial oxygen being dissociated from oxygen dimers to form two interstitial oxygen atoms; k.sub.bn is a dissociation rate constant of interstitial oxygen being dissociated from oxygen clusters of a cluster number n to form two interstitial oxygen atoms; n satisfies 3≤n≤M−1 in the above formula (3); and the formula (3) represents (M−3) simultaneous equations expressing change of [O.sub.n] where 3≤n≤M−1 with time, and in the third step, the thermal donor formation rate is calculated using the following formula (5) in which a minimum cluster number of the oxygen clusters that form donors is m, $\begin{array}{cc}\frac{d\left[\mathrm{TD}\right]}{\mathrm{dt}}=2\underset{n=m}{\overset{M}{.Math.}}\frac{d\left[O_n\right]}{\mathrm{dt}},& \left(5\right)\end{array}$ where [TD] represents a concentration of thermal donors TD.

2. The method of predicting a behavior of thermal donors, according to claim 1, wherein the combination rate coefficients k.sub.f1 and k.sub.f2 in the above formulae (1) to (4) are expressed by the following formulae (6) and (7):
k.sub.f1=4πr.sub.c D.sub.Oi  (6),
k.sub.f2=4πr.sub.c D.sub.O2  (7), where r.sub.c is a capture radius within which the interstitial oxygen O.sub.i and the oxygen dimers O.sub.2 are captured by oxygen clusters O.sub.n, D.sub.oi is a diffusion coefficient of the interstitial oxygen O.sub.i, and D.sub.O2 is a diffusion coefficient of the oxygen dimers O.sub.2.

3. The method of predicting a behavior of thermal donors, according to claim 2, wherein the diffusion coefficient D.sub.oiof interstitial oxygen O.sub.i in the above formula (6) has temperature dependency with a threshold value being 450° C.

4. The method of predicting a behavior of thermal donors, according to claim 3, wherein when the heat treatment temperature T is 450° C. or higher, the diffusion coefficient D.sub.oi is expressed by the following formula (8):
D.sub.Oi=α.sub.H.Math.exp(−β.sub.H/k.sub.BT).. .(8), where α.sub.H and β.sub.H are positive constants, k.sub.B is the Boltzmann constant, and T≥[K], when the heat treatment temperature T is lower than 450° C., the diffusion coefficient D.sub.oi is expressed by the following formula (9):
D.sub.oi=α.sub.L.Math.exp(−β.sub.L/k.sub.BT)  (9), where α.sub.L and β.sub.L are positive constants, kB is the Boltzmann constant, and T<723 [K], and α.sub.H>α.sub.L and β.sub.H>β.sub.L are satisfied.

5. The method of predicting a behavior of thermal donors, according to claim 4, wherein α.sub.H=0.13, α.sub.L=5.1×10.sup.−12, β.sub.H=2.53 [eV], β.sub.L=1.0 [eV] in the above formulae (8) and (9).

6. A method of evaluating a silicon wafer, comprising: a step of determining a concentration of thermal donors formed in the silicon wafer through heat treatment under a predetermined set of conditions using the method of predicting a behavior of thermal donors, according to claim 1; and a step of determining a predicted resistivity of the silicon wafer having been subjected to heat treatment under the predetermined set of conditions based on the thermal donor concentration.

7. A method of producing a silicon wafer comprising: a step of ascertaining a set of heat treatment conditions in a device process performed on the silicon wafer; a step of determining a predicted resistivity of the silicon wafer having been subjected to heat treatment under the set of heat treatment conditions for the device process using the method of evaluating a silicon wafer, according to claim 6; and a step of designing a target value of either an oxygen concentration or a resistivity of the silicon wafer before being subjected to the device process, based on the predicted resistivity.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) In the accompanying drawings:

(2) FIG. 1 is a graph illustrating the reproducibility of the thermal donor concentration during long-time low-temperature heat treatment, found by a formula obtained by improving a conventional technique;

(3) FIG. 2 is a graph illustrating the reproducibility of the thermal donor formation rate during short time low-temperature heat treatment, found by the above improved formula;

(4) FIG. 3 is a flow chart illustrating a method of predicting the formation behavior of thermal donors in a silicon wafer according to one embodiment of this disclosure; and

(5) FIG. 4 is a graph illustrating the reproducibility of the thermal donor formation rate during short time low-temperature heat treatment in Example 1;

(6) FIG. 5 is a graph illustrating the reproducibility of the thermal donor concentration during long time low-temperature heat treatment in Example 2;

(7) FIG. 6 is a graph illustrating the relationship between the oxygen concentration of a silicon wafer and the thermal donor formation rate at a heat treatment temperature of 450° C. in Example 3;

(8) FIG. 7 is a graph illustrating the relationship between the oxygen concentration of a silicon wafer and the thermal donor formation rate at a heat treatment temperature of 400° C. in Example 3;

(9) FIG. 8 is a graph illustrating the relationship between the oxygen concentration of a silicon wafer and the thermal donor formation rate at a heat treatment temperature of 350° C. in Example 3; and

(10) FIG. 9 is a graph illustrating an Arrhenius plot of the heat treatment temperature and the diffusion coefficient of interstitial oxygen in Example 3.

DETAILED DESCRIPTION

(11) Prior to the description of one embodiment of this disclosure, a bond-dissociation model of oxygen clusters used in this disclosure is described. Note that an oxygen cluster herein means a constitution of a plurality of interstitial oxygen atoms O.sub.i bound together in which the number of the oxygen atoms is two or more, and is expressed as an oxygen cluster O.sub.n using the number n of the oxygen atoms in the interstitial oxygen atoms O.sub.i. Hereinafter, the number of oxygen atoms in an oxygen cluster is referred to as a cluster number. Meanwhile, the maximum number of oxygen clusters that become donors is M (M is a natural number). Accordingly, the oxygen atom number (cluster number) n is an integer satisfying 2≤n≤M. Typically, when the cluster number of oxygen clusters is 4 or larger, the oxygen clusters are considered to become donors, and when the cluster number is excessively large, the oxygen clusters are considered to be electrically inactive or there is considered to be no oxygen cluster. This disclosure also follows this knowledge. Further, particularly when the cluster number is 2, the oxygen cluster is called an oxygen dimer, particularly when the cluster number is 3, the oxygen cluster is called an oxygen trimer. Oxygen dimers and oxygen trimers are considered electrically neutral.

(12) First, the inventors envisaged a bond-dissociation model in which the cluster number of oxygen clusters increases or decreases due to the bond dissociation of interstitial oxygen O.sub.i through the diffusion of interstitial oxygen O.sub.i. Here, the combination rate coefficient k.sub.f1 of when interstitial oxygen O.sub.i was bound to other interstitial oxygen or oxygen clusters was assumed to be fixed without depending on the cluster number of oxygen clusters after the combination. Meanwhile, the dissociation rate constant of when interstitial oxygen O.sub.i was dissociated from oxygen clusters O.sub.n was expressed as k.sub.bn, and the dissociation rate constant of when interstitial oxygen O.sub.i was dissociated from oxygen dimers O.sub.2 to form two interstitial oxygen atoms O.sub.i was designated k.sub.b2. They assumed that not all k.sub.b2, k.sub.b3, . . . , k.sub.bn, . . . , k.sub.bM would agree with each other. Based on this bond-dissociation model, the chemical equilibrium formula representing the model in which oxygen clusters O.sub.n are bound or dissociated through the diffusion of interstitial oxygen O.sub.i is as follows. Note that they assumed that interstitial oxygen O.sub.i was not dissociated from oxygen clusters of an oxygen cluster number of 13 or more, or even when some was dissociated, it would not be substantial dissociation. Further, the combination rate coefficient k.sub.f1 is dependent on the heat treatment temperature as is typically believed.

(13) $\begin{array}{cc}\begin{array}{c}\begin{array}{c}\begin{array}{c}\begin{array}{c}\begin{array}{c}{O}_i+O_i\underset{k_{b2}}{\stackrel{k_{f1}}{\rightleftarrows }}{O}_2\\ O_2+O_i\underset{k_{b3}}{\stackrel{k_{f1}}{\rightleftarrows }}{O}_3\end{array}\\ O_3+O_i\underset{k_{b4}}{\stackrel{k_{f1}}{\rightleftarrows }}{O}_4\end{array}\\ \mathrm{.Math.}\end{array}\\ O_{11}+O_i\underset{k_{b12}}{\stackrel{k_{f1}}{\rightleftarrows }}{O}_{12}\end{array}\\ O_{12}+O_i\begin{array}{c}\underset{.fwdarw.}{k_{f1}}\\ \end{array}{O}_{13}\end{array}& \left[1\right]\end{array}$

(14) Further, the inventors also envisaged a bonding model in which the cluster number of oxygen clusters increases by 2 due to the bonding of oxygen dimers O.sub.2 through the diffusion of the oxygen dimers O.sub.2. Here, the combination rate coefficient k.sub.f2 of when oxygen dimers O.sub.2 were bound to other interstitial oxygen or oxygen clusters was assumed to be fixed without depending on the cluster number of oxygen clusters after the combination. Based on this bonding model, the chemical equilibrium formula of oxygen clusters O.sub.n formed through the diffusion of oxygen dimers O.sub.2 is as follows. The combination rate coefficient k.sub.f2 is dependent on the heat treatment temperature as is typically believed.

(15) $\begin{array}{cc}\begin{array}{c}{O}_i+O_2\stackrel{k_{f2}}{.fwdarw.}{O}_3\\ O_2+O_2\stackrel{k_{f2}}{.fwdarw.}{O}_4\\ \mathrm{.Math.}\\ O_{10}+O_2\stackrel{k_{f2}}{.fwdarw.}{O}_{12}\\ O_{11}+O_2\stackrel{k_{f2}}{.fwdarw.}{O}_{13}\end{array}& \left[2\right]\end{array}$

(16) The reaction rate equations of interstitial oxygen, oxygen dimers, and oxygen clusters according to the above bond-dissociation model of interstitial oxygen O.sub.i and oxygen dimers O.sub.2 are expressed by the following formulae (1) to (4). In the formulae (1) to (4), [O.sub.1] and [O.sub.n] represent the concentrations of interstitial oxygen O.sub.i and oxygen clusters O.sub.n, respectively, and in particular, [O.sub.2] represents the concentration of oxygen dimers.

(17) $\begin{array}{cc}\frac{d\left[O_1\right]}{\mathrm{dt}}=-2{k_{f1}\left[O_1\right]}^2-k_{f1}\left[O_1\right]\left(\underset{n=2}{\overset{M-1}{.Math.}}\left[O_n\right]\right)+2k_{b2}\left[O_2\right]-k_{f2}\left[O_2\right]\left[O_1\right]+\underset{n=3}{\overset{M-1}{.Math.}}{k}_{\mathrm{bn}}\left[O_n\right]& \left(1\right)\\ \frac{d\left[O_2\right]}{\mathrm{dt}}={k_{f1}\left[O_1\right]}^2-k_{f1}\left[O_2\right]\left[O_1\right]-k_{b2}\left[O_2\right]+k_{b3}\left[O_3\right]-k_{f2}\left[O_2\right]\left(\left[O_1\right]+2\left[O_2\right]+\underset{n=3}{\overset{M-2}{.Math.}}\left[O_n\right]\right)& \left(2\right)\\ \frac{d\left[O_n\right]}{\mathrm{dt}}=k_{f1}\left[O_1\right]\left[O_{n-1}\right]-k_{f1}\left[O_1\right]\left[O_n\right]+k_{f2}\left[O_2\right]\left(\left[O_{n-2}\right]-\left[O_n\right]\right)-k_{\mathrm{bn}}\left[O_n\right]+k_{\mathrm{bn}+1}\left[O_{n+1}\right]& \left(3\right)\\ \frac{d\left[O_M\right]}{\mathrm{dt}}=k_{f1}\left[O_1\right]\left[O_{M-1}\right]-k_{f1}\left[O_1\right]\left[O_M\right]+k_{f2}\left[O_2\right]\left[O_{M-2}\right]-k_{\mathrm{bM}}\left[O_M\right]& \left(4\right)\end{array}$

(18) The following formula (10) holds since the concentration [TD] of thermal donors TD is the sum of the minimum cluster number of m and the maximum cluster number of M with which oxygen clusters O.sub.n become donors. The right-hand side coefficient of 2 in the following formula (10) is based on the assumption that one oxygen cluster having become a donor releases two electrons.

(19) $\begin{array}{cc}\left[\mathrm{TD}\right]=2\underset{n=m}{\overset{M}{.Math.}}\left[O_n\right]& \left(10\right)\end{array}$

(20) Accordingly, the formation rate of thermal donors TD formed in the silicon wafer by being subjected to heat treatment under a predetermined set of conditions follows the following formula (5) based on the above formulae (1) to (4).

(21) $\begin{array}{cc}\frac{d\left[\mathrm{TD}\right]}{\mathrm{dt}}=2\underset{n=m}{\overset{M}{.Math.}}\frac{d\left[O_n\right]}{\mathrm{dt}}& \left(5\right)\end{array}$

Method of Predicting Behavior of Thermal Donors in Silicon Wafer

(22) With the above in mind, a method of predicting the formation behavior of thermal donors in a silicon wafer according to one embodiment of this disclosure is described with reference to the flow chart of FIG. 3.

(23) A method of predicting the behavior of thermal donors formed due to oxygen generated when heat treatment is performed on a silicon wafer, according to one embodiment of this disclosure includes: a first step S10 of setting an initial oxygen concentration condition before performing the heat treatment on the silicon wafer for reaction rate equations based on both a bond-dissociation model of oxygen clusters O.sub.n associated with the diffusion of interstitial oxygen O.sub.i and a bonding model of oxygen clusters O.sub.n associated with the diffusion of oxygen dimers O.sub.2; a second step S20 of calculating the formation rate of oxygen clusters formed through the heat treatment using the reaction rate equations; and a third step S30 of calculating the formation rate of thermal donors formed through the heat treatment based on the formation rate of the oxygen clusters.

(24) Here, the reaction rate equations used in this embodiment are the formulae (1) to (4) mentioned above, and are reshown below:

(25) $\begin{array}{cc}\frac{d\left[O_1\right]}{\mathrm{dt}}=-2{k_{f1}\left[O_1\right]}^2-k_{f1}\left[O_1\right]\left(\underset{n=2}{\overset{M-1}{.Math.}}\left[O_n\right]\right)+2k_{b2}\left[O_2\right]-k_{f2}\left[O_2\right]\left[O_1\right]+\underset{n=3}{\overset{M-1}{.Math.}}{k}_{\mathrm{bn}}\left[O_n\right],& \left(1\right)\\ \frac{d\left[O_2\right]}{\mathrm{dt}}={k_{f1}\left[O_1\right]}^2-k_{f1}\left[O_2\right]\left[O_1\right]-k_{b2}\left[O_2\right]+k_{b3}\left[O_3\right]-k_{f2}\left[O_2\right]\left(\left[O_1\right]+2\left[O_2\right]+\underset{n=3}{\overset{M-2}{.Math.}}\left[O_n\right]\right),& \left(2\right)\\ \frac{d\left[O_n\right]}{\mathrm{dt}}=k_{f1}\left[O_1\right]\left[O_{n-1}\right]-k_{f1}\left[O_1\right]\left[O_n\right]+k_{f2}\left[O_2\right]\left(\left[O_{n-2}\right]-\left[O_n\right]\right)-k_{\mathrm{bn}}\left[O_n\right]+k_{\mathrm{bn}+1}\left[O_{n+1}\right],& \left(3\right)\\ \frac{d\left[O_M\right]}{\mathrm{dt}}=k_{f1}\left[O_1\right]\left[O_{M-1}\right]-k_{f1}\left[O_1\right]\left[O_M\right]+k_{f2}\left[O_2\right]\left[O_{M-2}\right]-k_{\mathrm{bM}}\left[O_M\right],& \left(4\right)\end{array}$
where t represents the time; M is the maximum number of oxygen clusters that become donors; [O.sub.1] and [O.sub.n] represent the concentrations of interstitial oxygen O.sub.i and oxygen cluster O.sub.n, respectively; k.sub.f1 is the combination rate coefficient of interstitial oxygen bound to either other interstitial oxygen O.sub.i or oxygen clusters, depending on a heat treatment temperature; k.sub.f2 is the combination rate coefficient of oxygen dimers bound to either other interstitial oxygen O.sub.i or oxygen clusters, depending on the heat treatment temperature; k.sub.b2 is the dissociation rate constant of interstitial oxygen being dissociated from oxygen dimers to form two interstitial oxygen atoms; k.sub.bn is the dissociation rate constant of interstitial oxygen being dissociated from oxygen clusters of a duster number n to form two interstitial oxygen atoms; n satisfies 3≤n≤M−1 in the above formula (3); and the formula (3) represents (M−3) simultaneous equations expressing change of [O.sub.n] where 3≤n≤M−1 with time.

(26) These steps will be sequentially described in detail below.

First Step

(27) In the first step S10, the initial oxygen concentration condition used in the above formulae (1) to (4) is set. The concentrations of [0.sub.1], [O.sub.2], . . . , [O.sub.n], . . . , and [O.sub.M] may be specifically set based on the oxygen concentration of the silicon wafer. Further, the oxygen concentration of the silicon wafer may be used as [O.sub.1] before low-temperature heat treatment assuming that all oxygen in the silicon wafer before being subjected to heat treatment is all interstitial oxygen O.sub.i. To simplify calculations, the initial oxygen concentration condition is preferably set as a condition for after performing donor killer heat treatment.

Second Step

(28) In the second step S20, the simultaneous differential equations of the formulae (1) to (4) may be numerically calculated in accordance with the conditions for heat treatment performed on the silicon wafer (heat treatment time t, heat treatment temperature T) under the initial oxygen concentration, set in the first step S10, thereby determining [O.sub.1], [O.sub.2], . . . , [O.sub.n], . . . , and [O.sub.M].

(29) Here, the combination rate coefficients k.sub.f1 and k.sub.f2 of interstitial oxygen O.sub.i and oxygen dimer O.sub.2 respectively may be handled as constants found by a regression analysis from test results, and preferably follows the following formulae (6) and (7):
k.sub.f1=4πr.sub.cD.sub.Oi  (6),
k.sub.f2=4πr.sub.cD.sub.O2  (7),

(30) where r.sub.c is a capture radius within which the interstitial oxygen O.sub.i and the oxygen dimers O.sub.2 are captured by oxygen clusters O.sub.n, D.sub.oi is a diffusion coefficient of the interstitial oxygen O.sub.i, and Do.sub.2 is a diffusion coefficient of the oxygen dimers O.sub.2). The capture radius r.sub.c has been reported in known literature and can be used. Further, the capture radius may be optimized to match the test results. The diffusion coefficients D.sub.oi and D.sub.O2 can be found in an Arrhenius plot.

(31) For example, in NPL 1 mentioned above, the capture radius r.sub.c is assumed to be 5 [angstrom] (0.5 nm). Further, for the diffusion coefficient D.sub.Oi of interstitial oxygen, D.sub.oi(T)=0.13×exp(−2.53 [eV]/k.sub.BT) [cm.sup.2/s] is disclosed in NPL 1 and PTL 1 mentioned above.

(32) Further, the dissociation rate constant k.sub.b2 of interstitial oxygen being dissociated from oxygen dimers to form two interstitial oxygen atoms and the dissociation rate constant k.sub.bn of interstitial oxygen being dissociated from oxygen clusters of a cluster number n can be found by a regression analysis using test results. According to the studies made by the inventors, the dissociation rate constant k.sub.bn is a value dependent on the temperature, and is approximately within the following range at 350° C. to 450° C.: 10.sup.−7≤k.sub.b2≤10.sup.−2, 10.sup.−6≤k.sub.b3≤10.sup.−2, 10.sup.−8≤k.sub.b4≤10.sup.−4, 10.sup.−9≤k.sub.b5≤10.sup.−5, 10.sup.−9≤k.sub.b6≤10.sup.−5, 10.sup.−9≤k.sub.b7≤10.sup.−4, 10.sup.−9≤k.sub.b8≤10.sup.−4, 10.sup.−8≤k.sub.b9≤10.sup.−4, 10.sup.−10≤k.sub.b10≤10.sup.−5, 10.sup.−10≤k.sub.b11≤10.sup.−5, 10.sup.−5≤k.sub.b12≤10.sup.−6 [s.sup.−1]. Further, the dissociation rate constant may be a function of an Arrhenius equation dependent on the temperature as in the formulae (8) and (9). It should readily be understood that the range of the dissociation rate constant (tendency) here is only an example, and when this embodiment is applied, various modifications are possible by optimization such that the range corresponds to that in the actual experimental results.

Third Step

(33) In the third step S30, the formation rate of thermal donors formed through heat treatment is calculated using the following formula (5) in which the minimum cluster number of m and the maximum cluster number of M allow oxygen clusters O.sub.n to become donors based on the formation rate of each oxygen cluster, calculated in the second step S20.

(34) $\begin{array}{cc}\frac{d\left[\mathrm{TD}\right]}{\mathrm{dt}}=2\underset{n=m}{\overset{M}{.Math.}}\frac{d\left[O_n\right]}{\mathrm{dt}}& \left(5\right)\end{array}$

(35) In general, the minimum cluster number m of an oxygen cluster that can become a donor is said to be 4; thus, preferably m=4. Further, according to a known document using Fourier-transform infrared spectroscopy (FTIR) (W. Gotz et al.: “Thermal donor formation and annihilation at temperatures above 500° C. in Czochralski-grown Si”; Journal of Applied Physics; Jul. 6, 1998; Vol. 84, No. 7, pp. 3561-3568), nine kinds of thermal donors are identified; thus, M=12 is preferred for the maximum cluster number M. It should readily be understood that m=4 and M=12 are mere examples, and when this embodiment is applied, various modifications are possible.

(36) Thus, through the first step S10, the second step S20, and the third step S30, the formation rate of thermal donors can be determined. Accordingly, the behavior of thermal donors formed while heat treatment is performed on a silicon wafer can be predicted more accurately than conventional techniques. Further, the prediction method according to this embodiment is implemented by calculations based on the bond-dissociation model of oxygen clusters described above, and thus can be applied to both short-time heat treatment and long-time heat treatment. In addition, based on the thermal donor formation rate, the thermal donor concentration (i.e., the amount of thermal donors) after heat treatment can also be determined. The reproducibility of experimental results obtained by this embodiment will be described in more detail using Examples 1 and 2 below.

(37) According to this embodiment, heat treatment at a lower temperature than donor killer heat treatment allows for predicting tendencies of the formation behavior of thermal donors from the diffusion coefficient D.sub.Oi of interstitial oxygen O.sub.i using a straight line approximation equation of a known Arrhenius plot regardless of the heat treatment temperature. As a straight line approximation of a known Arrhenius plot, for example, D.sub.oi(T)=0.13×exp(−2.53 [eV]/k.sub.BT) [cm.sup.2/s] is known. Here, as described in more detail using Example 3 below, the studies made by the inventors confirmed that the temperature dependence of the diffusion coefficient D.sub.oi of interstitial oxygen O.sub.i changed at a threshold of 450° C. In the light of this experimental result, the diffusion of interstitial oxygen O.sub.i is considered to be promoted (such a phenomenon is hereinafter referred to as “enhanced diffusion”) at temperatures lower than 450° C.

(38) Accordingly, in the light of the experimental results of Example 3 below, in order to predict the formation behavior of thermal donors more accurately, the diffusion coefficient D.sub.oi is preferably handled as having temperature dependence that changes at the threshold: a heat treatment temperature of 450° C. Thus, the diffusion coefficient D.sub.oi of interstitial oxygen O.sub.i in the formula (6) above preferably has temperature dependence that changes at a threshold of 450° C.

(39) Given these factors, when the heat treatment temperature T is 450° C. or higher, the diffusion coefficient D.sub.oi is expressed by the following formula (8):
D.sub.Oi=α.sub.H.Math.exp(−β.sub.H/k.sub.BT)  (8),

(40) where α.sub.H and β.sub.H are positive constants, k.sub.B is the Boltzmann constant, and T≥723 [K]),

(41) when the heat treatment temperature T is lower than 450° C., the diffusion coefficient D.sub.oi is expressed by the following formula (9):
D.sub.oi=α.sub.L.Math.exp(−β.sub.L/k.sub.BT)  (9),

(42) where α.sub.L and β.sub.L are positive constants, kB is the Boltzmann constant, and T<723 [K], and

(43) preferably α.sub.H<α.sub.L and β.sub.H>β.sub.L. In particular, in the above formulae (8) and (9), preferably α.sub.H=0.13, α.sub.L=5.1×10.sup.−12 and β.sub.H=2.53 [eV], β.sub.L=1.0 [eV].

(44) Using the enhanced diffusion model mentioned above, the formation behavior of thermal donors can be predicted more accurately.

(45) Specific aspects in this embodiment will be described below; however, these are mere examples.

(46) As the silicon wafer, a slice of a single crystal silicon ingot grown by the Czochralski process (CZ process) or the floating zone melting process (FZ process), cut with a wire saw or the like, can be used. The oxygen concentration of a silicon wafer formed by the CZ process (hereinafter “CZ wafer”) is higher than that formed by the FZ process and is susceptible to the change in resistivity due to the formation of thermal donors. For this reason, the prediction method of this embodiment is preferably used for a CZ wafer.

(47) The conductivity type of a silicon wafer to which the prediction method of this embodiment is applied may either be the p-type or the n-type. Further, the oxygen concentration of the silicon wafer is not particularly limited and is typically 4×10.sup.17 atoms/cm.sup.3 to 22×10.sup.17 atoms/cm.sup.3 (ASTM F121-1979). The prediction method of this embodiment is preferably applied to a silicon wafer having an oxygen concentration of 10×10.sup.17 atoms/cm.sup.3 (ASTM F121-1979) or higher in which the thermal donors have significant effects.

(48) Further, heat treatment conditions in this embodiment are not particularly limited; the method is effective in predicting the behavior of thermal donors formed by heat treatment for a significantly short time of approximately several minutes to several hours, and the method is also effective in predicting the behavior of thermal donors formed by heat treatment for a long time of more than 100 hours. Specifically, this embodiment is preferably applied when the heat treatment time is 1 minute or more and 50 hours or less, and is more preferably applied when the heat treatment time is 20 hours or less. On the other hand, this embodiment is preferably applied when the heat treatment time is 100 hours or more, and this embodiment is more preferably applied when the heat treatment time is 200 hours or more. The heat treatment temperature is also not particularly limited as long as the heat treatment is not donor killer heat treatment; for example, this embodiment is preferably applied to heat treatment at 300° C. or more and 600° C. or less, and is particularly preferably applied to heat treatment at a temperature lower than 450° C. Further, in calculating the thermal donor formation rate or the thermal donor concentration after heat treatment in this embodiment, the heat treatment temperature need not be constant. Heat treatment conditions simulating an actual device process can be applied.

(49) Note that the above bond-dissociation model of oxygen clusters need not necessarily be theoretically true, and this embodiment predicts the thermal donor formation behavior in a silicon wafer according to the formulae (1) to (5) above.

(50) Further, the calculations of the formation rate and the amount of oxygen clusters formed under a predetermined set of heat treatment conditions (i.e., oxygen cluster concentration after heat treatment) are inextricably linked to each other; similarly, the calculations of the formation rate and the formation amount (i.e., thermal donor concentration) of thermal donors are inextricably linked to each other. Each of these is one aspect in predicting the thermal donor formation behavior. Accordingly, calculating the concentration of oxygen clusters or thermal donors after heat treatment is equivalent to calculating the formation rate of the oxygen clusters.

Method of Evaluating Silicon Wafer

(51) Further, a silicon wafer can be evaluated using the embodiment of a method of predicting the formation behavior of thermal donors in a silicon wafer. First, a step of determining the concentration of thermal donors formed in a silicon wafer through heat treatment under a predetermined set of conditions is performed according to the embodiment of a method of predicting the formation behavior of thermal donors. For the predetermined set of conditions, heat treatment history on the silicon wafer in the device process is preferably used. When high-temperature heat treatment corresponding to donor killer heat treatment is included, only the heat treatment history after the high-temperature heat treatment may be used. Based on the thermal donor concentration found, a step of determining the predicted resistivity of the silicon wafer having been subjected to the heat treatment under the predetermined set of conditions is performed. The resistivity can be determined using Irvin curves based on the concentration of the thermal donors formed. This method of evaluating a silicon wafer allows for accurately evaluating whether or not the resistivity of a silicon wafer which is subjected to heat treatment under a predetermined set of conditions meets a specific specification.

Method of Producing Silicon Wafer

(52) Further, it is also preferred to produce a silicon wafer using the above evaluation method. First, a step of ascertaining a set of heat treatment conditions in a device process performed on the silicon wafer is performed. When high-temperature heat treatment corresponding to donor killer heat treatment is included, the inclusion and heat treatment conditions after the high-temperature heat treatment may be ascertained, whereas when high-temperature heat treatment is not included, the entire thermal history is preferably ascertained. Whereupon, a step of determining the predicted resistivity of the silicon wafer having been subjected to heat treatment under the set of heat treatment conditions for the device process is performed using the method of evaluating a silicon wafer. Next, based on the predicted resistivity determined, a step of designing a target value of either the oxygen concentration or the resistivity of the silicon wafer before being subjected to the device process is performed, thereby producing a silicon wafer according to this design. With the use of a silicon wafer fabricated using the method of producing a wafer, of which resistivity after the device process is considered, detrimental effects on the device characteristics due to the formation of thermal donors can be reduced.

EXAMPLE 1

Prediction Accuracy when Short-Time Heat Treatment is Performed

(53) First, to grasp the reproducibility of the method of predicting the thermal donor formation behavior in a silicon wafer according to this disclosure, the inventors compared the experimental value and the calculated value of the thermal donor formation rate when short-time low-temperature heat treatment was performed. The experimental value of the thermal donor formation rate was determined first in the following manner.

(54) A (100) p-type silicon single crystal ingot having a diameter of 300 mm was grown by the CZ process. After slicing the single crystal silicon ingot to form silicon wafers, the oxygen concentration of each silicon wafer was measured by Fourier-transform infrared spectroscopy. The oxygen concentration of the silicon wafers varied from 4×10.sup.17 [cm.sup.−3] to 22×10.sup.17 [cm.sup.−3] (ASTM F121-1979, hereinafter the same applies to the oxygen concentration) depending on the growth conditions and slicing positions of the single crystal silicon ingot. Of those, a silicon wafer with an oxygen concentration of 11×10.sup.17 [cm.sup.−3] was used. In order to eliminate thermal donors formed in the crystal growth of the silicon wafer, donor killer treatment was performed in a nitrogen atmosphere at 700° C. for 15 min. After that, low-temperature heat treatment was performed in a nitrogen atmosphere at 450° C. for 2 hours to 40 hours, thereby generating thermal donors in the silicon wafer. The specific resistance of each silicon wafer was measured according to a specific resistance measurement method using a four-point probe array according to JIS H 0602 (1995). The carrier concentration was measured from Irvin curves based on the measurement result of the specific resistance and the measurement result of the specific resistance before low-temperature heat treatment. Further, based on the carrier concentrations before and after low-temperature heat treatment by which thermal donors were formed, the amount of thermal donors formed was determined. Further, the thermal donor formation rate was found from the relationship between the heat treatment time and the amount of thermal donors formed (thermal donor concentration). The measured thermal donor formation rate is given in FIG. 4.

(55) Next, the thermal donor formation rate of a silicon wafer equivalent to one in the above experiment, which had been subjected to low-temperature heat treatment was calculated using the formulae (1) to (7) according to this disclosure. Oxygen after donor killer treatment was all assumed to be interstitial oxygen O.sub.i: [O.sub.i]=11×10.sup.17 [cm.sup.−3]. Note that as the diffusion coefficient of interstitial oxygen O.sub.i, D.sub.oi(T)=0.13×exp(−2.53 [/k.sub.BT) cm.sup.2/s] in the known document was used. Further, as the diffusion coefficient D.sub.O2 of oxygen dimers D.sub.O2(T)=1.7×exp(−1.9 [eV]/k.sub.BT) [cm.sup.2/s] was used. Further, the capture radius r.sub.c was set to 7 [angstrom] (0.7 nm). The dissociation rate constants k.sub.b2, k.sub.b3, . . . , k.sub.bn, and k.sub.bM were optimized by a regression analysis based on the above experimental values. Further, the minimum cluster number with which oxygen clusters become donors was set to 4 (m=4), and the maximum cluster number with which oxygen clusters become donors was set to 12 (M=12). The result of numerical calculations is given as Example 1 in FIG. 4.

(56) For the comparison with Example 1, the calculation result of the thermal donor formation rate according to Formula B mentioned above is also given in FIG. 4.

(57) As illustrated in FIG. 4, it is experimentally confirmed that the thermal donor formation rate increases for the first approximately 10 hours of heat treatment and decreases after that. This decrease was found to be accurately reproduced in Example 1. On the other hand, using Formula B described above, the formation behavior of thermal donors in such a heat treatment for an extremely short time cannot be reproduced.

EXAMPLE 2

Prediction Accuracy when Long-Time Heat Treatment is Performed

(58) Next, to grasp the reproducibility of the method of predicting the thermal donor formation behavior in a silicon wafer according to this disclosure, the experimental value and the calculated value of the thermal donor concentration when long-time low-temperature heat treatment was performed were compared. For the experimental value, of the concentrations of thermal donors formed by low-temperature heat treatment disclosed in NPL 2 (Y. Kamiura et al.), the value of when the oxygen concentration of the silicon wafer was 13×10.sup.17 [cm.sup.−3] was used. FIG. 5 depicts the experimental values. The heat treatment temperature was 450° C., and the heat treatment time was 10 hours to more than 1000 hours.

(59) On the other hand, the thermal donor concentration of a silicon wafer equivalent to one in the above experiment, which had been subjected to low-temperature heat treatment was calculated using the formulae (1) to (7) and (10) according to this disclosure. The parameters used in Example 2 were the same as those in Example 1 except for the oxygen concentration of the silicon wafer and heat treatment time. The result of numerical calculations is given as Example 2 in FIG. 5.

(60) For the comparison with Example 1, the calculation result of the thermal donor formation rate according to Formula B mentioned above is also given in FIG. 5.

(61) It was confirmed that a reduction of the thermal donor concentration by low-temperature heat treatment for more than 400 hours was reproduced in Example 2. On the other hand, using Formula B described above, the formation behavior of thermal donors in such a heat treatment for a long time cannot be reproduced.

EXAMPLE 3

Temperature Dependence of Diffusion Coefficient

(62) To study the oxygen concentration dependence of the thermal donor formation rate, the inventors conducted the following experiment. The experimental value of the thermal donor formation was determined first in the following manner.

(63) The silicon wafers having different oxygen concentrations (4×10.sup.17 to 22×10.sup.17 [cm.sup.−3]) formed in Example 1 were subjected to donor killer treatment in a nitrogen atmosphere at 700° C. for 15 min. After that, thermal donor formation heat treatment was performed in a nitrogen atmosphere at 450° C., 400° C., and 350° C., thereby forming thermal donors. The heat treatment time was 32 hours at 450° C., 32 hours at 400° C., and 80 hours at 350° C. The thermal donor formation rate was determined similarly as in Example 1. Experimental values of the thermal donor formation rate at temperatures of 450° C., 400° C., and 350° C. are given in FIG. 6, FIG. 7, and FIG. 8, respectively.

(64) First, the thermal donor formation rate was calculated using the formulae (1) to (7) according to this disclosure as in Example 1. The parameters used for this calculation were the same as those in Example 1 except for the oxygen concentration of the silicon wafer and heat treatment time. The results obtained by this calculation for 450° C., 400° C., and 350° C. are given as Example 3, Example 4A, and Example 5A in FIG. 6, FIG. and FIG. 8, respectively.

(65) Also when the results of Examples 4A and 5A are used, the tendency of thermal donors to increase or decrease in their formation behavior can be grasped to a certain degree. However, FIG. 6, FIG. 7, and FIG. 8 suggest that the enhanced diffusion of interstitial atoms increased the diffusion coefficient thereof at a temperature lower than 450° C. Based on these experimental results, the enhancement corresponding to the diffusion coefficient illustrated in FIG. 9 was assumed using the formulae (8) and (9) according to this disclosure. In FIG. 9, α.sub.H=0.13, α.sub.L=5.1×10.sup.−12, β.sub.H=2.53 [eV], and β.sub.L=1.0 [eV]. The graphs in FIGS. 7 and 8 were obtained by recalculating the thermal donor formation rate of when the heat treatment temperature was 400° C. and 350° C. (Example 413 and Example 5B, respectively) considering the enhanced diffusion.

(66) FIGS. 7 and 8 demonstrate that the experimental results were well reproduced by the enhanced diffusion model of interstitial oxygen at a temperature lower than 450° C. At a temperature lower than 450° C., interstitial silicon and interstitial oxygen form a complex to undergo enhanced diffusion; however, since its bonding is instable, they would be dissociated at high temperatures.

INDUSTRIAL APPLICABILITY

(67) This disclosure provides a method of accurately predicting the thermal donor formation behavior in a silicon wafer that can be applied to both short-time heat treatment and long-time heat treatment, a method of evaluating a silicon wafer using the prediction method, and a method of producing a silicon wafer using the evaluation method.