JPS59108119A - On-line measuring and controlling system of furnace temperature - Google Patents

Abstract

PURPOSE:To make the internal temperature of an object to be heated in a furnace close to an objective temperature by estimating the furnace temperature from picture information obtained from ITVs and inputting the estimated value to a control model. CONSTITUTION:The ITVs 107-111 are fitted to windows 1030, 103-106 of the furnace 101. The ITVs 107-111 sends the distribution data 112-116 of the furnace temperature to a picture processor 117, which estimates the external temperature of the object in the furnace and applies the estimated temperature information 118 to a temperature conversion calculating device 119. The device 119 applies the estimated temperature information 118 to a control model computer 121 as the temperature input information 120 necessary for a control model. The control model computer 121 calculates the temperature inside the object to be heated by using the temperature distribution estimating model of the object 102 to be heated and controls the flow rate of feeding gas, the heating temperature of a heater, etc. by a closed loop. Consequently, the dispersion of products can be reduced.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to control systems for semiconductor oxidation/diffusion furnaces and CVD furnaces, steel blast furnaces, glass melting furnaces, paper digesters, chemical-related reactors, etc. In particular, the present invention relates to an online measurement method for the internal temperature distribution of a heated material placed in a furnace and a furnace control method.

[Prior art]

Conventionally, in order to control the temperature t1 distribution inside the furnace, methods such as (1) control based on outer wall temperature prediction (control by visual inspection of municipal blast furnaces, etc.) have been used.

(i+ is a method that measures the temperature of the outer wall of the blast furnace, predicts the temperature distribution inside the furnace based on this actual measurement value, and performs the desired control. This is a method of predicting the distribution in the reactor and performing the desired control, but these methods rely on intuition and know-how based on past experience, making precise furnace control impossible.

For example, when controlling the temperature of a diffusion furnace to uniformly diffuse the entire wafer at a temperature determined by the design, conventional diffusion furnaces do not maintain the temperature of the entire furnace wall at a constant temperature. Although it was controlled, the wafer temperature was hardly known because there was no method to measure the temperature of the wafer.

If a temperature sensor is installed near the wafer, the injected gas will be in a turbulent state, which will impair the uniformity of the wafer temperature, so it is necessary to avoid using the sensor to measure the temperature in the reactor core. For these reasons, there is a need to develop a model that can estimate the wafer temperature inside the furnace from the furnace wall temperature. The conventional method of heating the furnace wall of a diffusion furnace to a constant temperature has the disadvantage that the temperature varies depending on the position of the wafer, resulting in variations in the characteristics of the final semiconductor and resulting in a reduction in yield. .

[Purpose of the invention]

The purpose of the present invention is to provide an online measurement method for the temperature distribution inside the furnace and a method for precisely controlling the furnace so that the internal temperature distribution of the 7'JDu object in the furnace can be operated close to the target temperature distribution. Our goal is to provide the following.

[Summary of the invention]

Because the temperature inside the furnace is high (for example, 1000t:'),
Due to issues such as durability, regular cell phones (such as thermocouples, etc.)
cannot be used for on-the-in-furnace measurements.

Therefore, in the present invention, a window (similar to the peephole that is currently installed) is provided at an appropriate position of the furnace, several ITVs (industrial televisions) are installed there, and images from the ITVs are installed. Based on the information, the temperature inside the furnace is estimated using an image processing device, and this estimated value is input into the control model to bring the internal temperature of the heated object in the furnace closer to the target temperature. There are characteristics.

[Embodiments of the invention]

FIG. 1 shows the structure of a first embodiment of a semiconductor oxidation/diffusion furnace according to the present invention. In FIG. 1, windows 1030, 103 to 106 (the number and shape of the windows vary depending on the thickness and control method of the furnace) are provided at appropriate locations in the furnace 101,
ITVs (industrial televisions) 107 to 111 are installed at the positions of these windows. This ITV provides temperature distribution information inside the furnace.
2 to 116 are sent to the image processing device 117. The image processing device 117 has a high-speed calculation function such as an ultra-high-speed pipeline processor, a large-capacity refresh memory function, and the ability to perform advanced image processing in real time. This device 117 estimates the external temperature of the object in the furnace (however, the space visible from the ITV) by analyzing its thermal spectrum, and inputs the temperature estimation information 118 to the temperature conversion calculation device 119. here,
Temperature estimation information from each ITV is used as input information necessary for the control model (for example, temperature distribution information of each ITV is averaged and averaged temperature/ITV). This temperature input information 120
is input into the control model calculation device 121. Here, we use a heated object (Ueno 102) temperature distribution estimation model,
Using the input temperature input information 120 as a boundary value (astronomically measured value), the in-plane temperature distribution of the wafers of all the wafers 102 in the furnace is calculated. The calculation method models thermofluid equations for wafer temperature, gas temperature, and gas flow rate, and uses temperature input information 120 as a boundary value, and also sets the inflow gas flow rate 12 at that time as a boundary value.
2 is measured from the gas flow meter 123, and furnace wall temperatures 124 to 126 are measured from thermometers 127 to 129, and based on these data, the temperature distribution within each wafer surface is estimated and calculated. At this time, the following evaluation formula % formula % (1) Where, i: Wafer A (i=l, 2. Old...
, N)T: Wafer in-plane target temperature 'ft(r): Temperature at the point of wafer position vector r αI(r): In the weighting coefficient, J is J-ΔJ<J<J+ΔJ... Song (2) Here, ΔJ: It is determined whether the evaluation is within the evaluation tolerance range. If it is, the current control will continue. If not, perform optimal calculation to minimize J in equation (1), gas control target value 130, heater heating target value 131 to 1.
Find 33. These target values are input to feedback control devices 134-137, known as PID control devices, to control the furnace on-line.

Next, based on a second embodiment of the present invention, it is possible to improve the drawback that when the temperature inside the heating furnace cannot be measured, the object to be heated is heated unevenly by simply keeping the furnace wall temperature constant. In order to achieve this, details of the procedure for controlling the temperature of the heated furnace wall to be heated at a uniform temperature by, for example, adjusting the temperature of the heated furnace wall using a heater installed on the furnace wall, based on a model that estimates the internal state of the furnace. Explain.

FIG. 2 shows the configuration of a second embodiment of the diffusion furnace according to the present invention,
An example of the connection relationship with the temperature control device is shown below. In FIG. 2, the target diffusion furnace 101 is a cylindrical tube, and the wafer 102 is placed on a bow 1-2020 in the center, transported into the furnace 101, diffused, and exited the furnace 201 again. The problem is estimating the temperature distribution inside the furnace in the direction of the furnace axis (Z-axis) during diffusion. Since the wafers in the center of the boat are heated by direct radiation from the wafers on both sides and by direct radiation from the furnace walls, it is assumed that the wafers in the center will be hotter than the wafers at both ends. Therefore, the temperature control of the diffusion furnace for diffusing the wafer at a uniform temperature is performed as follows. First, as an initial input, for example, the design (target) temperature T of sea urchin... is input into the optimum calculation device 206, and the wafer is heated to
, 204 and 205 to determine the furnace wall temperatures of the three symmetrical zones to achieve uniform heating. When the furnace wall temperature is determined, the difference between the furnace wall temperature measurement result by the sensor 208 and the furnace wall temperature determined by the optimum calculation device 6 is calculated by the subtracter 200, and based on this difference, the PID co/controller 207, adjust the power supply 209, and connect the heaters 203 and 2
04 and 205 to a predetermined temperature.

Next, the procedure for setting the furnace wall temperature will be explained. The temperature of the i-th wafer is TI, the temperature of the two adjacent circles on both sides of the i-th is Tj (j=i±1), the furnace wall temperature is TF, the thermal conductivity of silicon is KW, and the absorption rate is a.

The view factor between wafers is G. , the view factor of Ueno and the furnace wall is F. Then, the i-th wafer temperature T1 can be obtained by solving the following thermal equation.

r') dAj+afTp(r') Fo (r*
r') dF・・・・・・・・・(1n (In the 1n equation, the second term represents the radiation from the wafers on both sides (their surface area is Aj), and the third term is the furnace wall (
The following approximation is made to the third term representing the radiation from the surface area of which is F. Wafer temperature Tj(r't'
) is a function for position r', and r') is a function that takes the maximum value when r = r', and for other r's, it is inversely proportional to 1r-r'l to the 4th power. This approximation is possible because it becomes small. For such an approximation, equation (1) can be written as follows.

+aTF'f(r)+KW(Tt)Vr2T+ ---
--- (2) (2) Expression 1. If it is divided into the middle part Ueno, it will be as follows.

KW(TI)VI'TI+KTF' (i=2.
3・−・・・・、N） ・・・・・・・・・・・・ （
3) Here, KTF' is a term representing radiation from the furnace wall. Now, we will focus on the wafer temperature in a steady state and omit the diffusion term to investigate temperature fluctuations between wafers.

-2Ti' + aTF'
=02T+'+a ('r'l-,+T
'l+1) g(r)+=0(i=2.3...
・,N) -2TN'+aTF' =:Q・
・・・・・・・・・(4) Distance between 1st and Nth wafers (#Boat length LB)
When constant, the discontinuous wafer position i in the Z-axis direction is
As N increases, the continuous amount Z=iΔZ(ΔZ=LB
/N). Therefore, T'+- and T41+ are expanded as follows.

θT θ2T = T' + 4T3-+ 2T3- ・Curved surface (6) θZ
az2 Substituting equations (5) and (6) into the second equation of equation (4), taking the second derivative with respect to Z, and rearranging K I as a temperature-dependent non-magnetic small constant, we can approximately obtain the following equation. It can be written as follows.

The first of equation (4). By using the boundary condition equation (3), the solution to equation (7) is + (e 1) e −' ) + u'Tr ・
...(8) LB. FIG. 3 shows the general state of the temperature distribution in the Z-axis direction expressed by equation (8). In this way, the wafers near both ends have a lower temperature than the wafers in the middle. However, the method of controlling the temperature equalization between wafers uses the furnace wall temperature TF as the only manipulated variable, and uses a control method that minimizes the temperature difference between the wafers by making both ends of the furnace wall temperature higher than the middle part. That's a thing. In order to realize this algorithm, we introduce the following evaluation function.

Here, T is the design temperature of the wafer. Now, assuming that the temperature of the diffusion furnace is controlled by a symmetrical three-zone system, T
(Z) becomes as follows.

Substituting equations (8) and (10) into equation (9), TU, T
By varying J with respect to L and setting it as O, we obtain the following equation.

Here, 2kLB -e ) + 2AB/= 2 'TF 2T Y' = (--) (Aek1''-Be-kLB3k k (L = -1) - Li Do Z) ) - Ae
+13e A22kt B2-kt X″-1 (e'k(LB'-t)-e) 10-(e2k
2 to −e 2k””−Z))+2. ABLa -4kBt2
'TF 2T-1-Be y// -(, (Aek(LB-1) -
k(LB-1)3kkl-kl-Ae+Be) Furthermore. FIG. 4 shows the general appearance of Tt and Tu.

The furnace wall temperature in the three zones was determined according to equation (11) so that the wafers could be heated uniformly, but what must be noted here is that by changing the furnace wall temperature, equation (7) can be used. The right-hand side of is a change. Therefore, returning to equation (8) again (
The optimal furnace wall temperatures TU and 'l'L can be obtained by repeating the algorithm of dividing To and TL that minimizes the evaluation function of equation 9). FIG. 5 shows the above algorithm.

In each step in FIG. 5, the following processing is performed.

Step 401: As an initial input, for example, a target design temperature T is inputted to the optimum calculation device 206.

Step 402: (Calculate the inter-wafer temperature distribution using the 8J formula.

Step 403: Calculate the temperatures To and Tt of each of the three symmetrical zones from equation (11).

Step 404: Using To and TL, calculate T(Z) again using equation (8).

Step 405: Minimize the evaluation function Tu.

Determine whether Tb is true (Y) set to the source. 1. If Th(N) Return to step 402 and use the algorithm Repeat rhythm.

〔Effect of the invention〕

According to the present invention, the external temperature of the 7JLl heated object in the furnace is clarified by image processing technology, and furthermore, using this temperature estimate, the internal temperature of the heated object can be estimated by a model. In order to bring internal protection closer to the target value,
Since the set values for input gas flow rate, heater galvanic temperature, etc. can be determined, furnace control, which previously relied on intuition and know-how, can be changed to closed-loop control based on online measurement, which is expected to have a significant effect on yield improvement. can. According to the method shown in the second embodiment, in a furnace wall temperature control method that uniformly heats a wafer in a diffusion furnace, a model created based on a heat equation and an evaluation function are used to calculate the temperature of the furnace wall. Since the furnace wall temperature is controlled so that the evaluation function is minimized, the optimal furnace wall temperature can be determined, making it possible to reduce variations in semiconductor characteristics and improve yield. This makes it possible to stably manufacture high-performance semiconductors, and the economic effects are significant.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a first embodiment of a diffusion furnace according to the present invention;
FIG. 2 is a block diagram of a second embodiment of the diffusion furnace according to the present invention;
Figure 3 shows the temperature distribution between wafers when the furnace wall is heated uniformly, and Figure 4 shows the temperature distribution of the furnace wall using the three-zone method.
FIG. 5 shows a flowchart of the control algorithm. Figure 3 Z Figure 4 Z (J-Lβ/3) '-1y s diagram

Claims (1)

[Scope of Claims] 1. In a target follow-up control method for the internal temperature of the object to be heated in the cutting edge thermal furnace, based on thermal spectrum information obtained from a phantom hand installed in the viewing window of the heating furnace. Relaxation in
Estimates the external temperature of the object, and estimates the internal temperature of the object by using the estimated external temperature, the furnace wall temperature measured by a temperature sensor installed on the furnace wall, information on waste inflow, etc. The control set values for the heater, gas inflow rate, etc. are determined based on the estimated value and the predetermined grade of the object to be heated so as to approach the target temperature. An online furnace temperature measurement and control method. 2. In the target additional value control method for the internal temperature of the object to be heated in the heating furnace, the furnace wall of the heating furnace is divided into zones in the direction of the furnace axis, and the furnace axis of the object to be heated is determined from the temperature of the furnace wall. A furnace wall temperature is determined based on a nonlinear model equation that estimates a directional temperature distribution and an evaluation function that evaluates the difference between the estimated value and a target value, so that the evaluation function is minimized. Temperature online measurement and control method.