JPH01282620A - Temperature controller for semiconductor processing furnace - Google Patents

Abstract

PURPOSE:To accurately control the in-furnace temperature to a furnace temperature target value without receiving the influence of a heater temperature detection error even at the time of a present cooling operation by preparing a heater temperature target value pattem at the time of the cooling operation while the in-furnace temperature control by a pre-cooling operation is executed. CONSTITUTION:When an in-furnace temperature is set to fall the temperature to a necessary furnace temperature value, the pre-cooling operation to execute the in-furnace temperature control to a furnace target value with the output value of an in-furnace temperature detector 28 directly as a feedback signal is performed. During the execution of the pre-cooling operation, the actual result data of the output value of a heater temperature detector 26 are picked up, and the heater temperature target value pattern at the time of a cooling operation is prepared based the actual result data. At the present cooling operation time, the temperature control of a heater 25 is executed with the heater temperature target value pattern as a heater temperature target value and with the heater temperature detecting value as a feedback signal. Thus, at the time of the present cooling operation, the in-furnace temperature can be accurately controlled without receiving the influence of the heater temperature detecting error.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to a temperature control device for a semiconductor processing furnace, such as a semiconductor diffusion furnace.

(Conventional Technique #I) Semiconductor processing furnaces such as semiconductor diffusion furnaces are equipped with the ability to quickly raise and lower the temperature in order to increase the productivity of semiconductor devices. ) that can be set to the required temperature in a short time.

FIG. 6 shows an example of the furnace body structure of such a conventional semiconductor diffusion furnace. In the figure, 1 is a heat insulating material, and heaters 2.3.4.5 as heating means are wound inside the heat insulating material, divided into 4 zones, for example, and each heater 2.3.4.
Heater temperature detectors 6, 7, 8, and 9 are attached to 5, respectively. A quartz tube (reaction tube) 11 is arranged inside these heaters 2.3.4.5, and a boat loader 12 is arranged below it. 13 is a boat pedestal, and 14 is a boat for loading wafers 15 to be processed.

21 is a quartz bar, and the quartz bar 21 is equipped with a furnace temperature detector 1.
6.17.18.19 are attached, and are inserted into the quartz tube 11 when measuring the furnace temperature near the wafer 15, as will be described later.

The signal line for the heater temperature detection value from the heater temperature detector 6.7.8.9 is connected to a temperature regulator (not shown), and a required temperature target value is given to this temperature regulator. The heater output of each zone is controlled using the heater temperature detected by each heater temperature detector 6.7.8.9 as a feedback signal, and the furnace temperature is set to a required temperature.

Further, a blower pipe 22 is connected to the upper opening of the heat insulating material 1, and a blower 23 as a forced cooling means for lowering the temperature inside the furnace is attached to the blower pipe 22. 24 is a cooler.

During forced air cooling operation, the blower 23 is operated, and the cooling air drawn from the lower part of the furnace body passes between the quartz tube 11 and each heater 2.3.4.5, and is drawn out from the upper part of the furnace body. The blower 2 is then connected via the blower piping 22 and the cooler 24.
3, the furnace temperature is lowered to the required temperature.

As described above, the main purpose of controlling the furnace temperature of the semiconductor diffusion furnace by heating or cooling is to equalize the temperature within the quartz tube 11 into which the object to be processed is loaded at a required temperature. In other words, the furnace temperature in the quartz tube 11 is precisely controlled to the required temperature, and the furnace temperature in each zone detected by the furnace temperature detectors 16, 17, 19 inserted in the quartz tube 11 is The goal is to suppress variations between zones. However, the furnace temperature detector 1 attached to the quartz bar 21
If 6.17.18.19 is always installed in the furnace, it will interfere with the insertion of the wafer 15. For this reason, the furnace temperature detectors 16, 17, 18, 19 are usually
5 is not inserted, it is inserted into the furnace and used for off-line data collection.

Therefore, as mentioned above, actual furnace temperature control is performed using the heater temperature detector 6.7.8.9, which has a strong correlation with the furnace temperature.
This is done using the detected heater temperature as a feedback signal.

This control method is also called auto-profile control, and uses correlation data between the heater temperature detection value and the furnace temperature in a steady state to create an offline conversion model from the heater temperature detection value to the furnace temperature. Using this conversion model, the furnace temperature is indirectly predicted from the heater temperature detection value, and this is used as a feedback signal to control the furnace temperature.

Incidentally, when collecting offline data as described above, it is possible to insert a dummy wafer 8 instead of the product wafer 15 to collect data in a state as close to online as possible.

(Problems to be Solved by the Invention) Conventional furnace temperature control in semiconductor diffusion furnaces was performed using the heater temperature detected by the heater temperature detector 6.7.8.9 as a feedback signal, which resulted in poor control accuracy. , No problem occurs during heating and steady operation, but during forced air cooling operation using the blower 23, cooling air flows between the quartz tube 11 and each heater 2.3.4.5, so the heater temperature detector 6.7.8.9 are agitated by the air and output a detected temperature value that is lower than the actual heater temperature.

For this reason, the heater temperature is evaluated to be low and fed back, so when controlling the furnace temperature, the actual heater temperature, and ultimately the furnace temperature, are controlled to be higher than the temperature target value, resulting in a decrease in control viscosity. There was a problem with this.

Furthermore, since the cooling air is gradually warmed from the time it is drawn into the furnace until it is drawn out of the furnace, the magnitude of the heater temperature detection error described above differs depending on the zone.

For this reason, in addition to the above-mentioned decrease in control accuracy, there is a problem in that the temperature difference in the furnace between zones increases during forced air cooling operation.

Further, there is a problem that the above-mentioned decrease in control accuracy and the like may adversely affect the performance of the semiconductor device subjected to the diffusion process, and may also cause a decrease in yield.

This invention was made based on the above circumstances, and it is possible to accurately control the furnace temperature to a required temperature target value without being affected by heater temperature detection errors during cooling operation, and also to achieve zone mutual control. An object of the present invention is to provide a temperature υ control device for a semiconductor processing furnace that can minimize the temperature difference in the furnace between the two.

[Structure of the Invention] (Means for Solving the Problems) In order to solve the above problems, the present invention provides a method for controlling the temperature inside a semiconductor processing furnace that is equipped with a heater that raises the temperature inside the furnace and a cooling means that lowers the temperature. A temperature control device that controls the furnace temperature to a desired furnace temperature target value using a detected value of the furnace temperature as a feedback signal during a pre-cooling operation to set the furnace temperature to a desired furnace temperature target value. and a temperature control means 1, which collects actual data of the heater temperature detection value of the heater section during the implementation of furnace temperature control by the first temperature control means, and determines a heater temperature target value pattern during cooling operation from this actual data. a second temperature control means for controlling the heater temperature during cooling operation by using the heater temperature target value pattern created by the calculation means as the heater temperature target value and using the detected heater temperature value as a feedback signal; The gist is to have the following.

(Function) In the above configuration, when setting the temperature inside the furnace to fall to the required furnace temperature target value, a furnace temperature detector is installed in the furnace as a preliminary operation when the workpiece to be treated is not inserted. The output of this furnace temperature detector is used as a direct feedback signal, and forced air cooling operation is performed while controlling the furnace temperature to the target value using this feedback signal. At this time, since the furnace temperature detector is located inside the furnace, such as inside a quartz tube, it is not stirred up by cooling air or the like, and detection errors that occur with heater temperature detectors do not occur. Therefore, during the pre-cooling operation, the furnace temperature is accurately controlled to the furnace temperature target value. Then, while the furnace temperature is being controlled by this pre-cooling operation, the actual data of the heater temperature detection value output from the heater temperature detector is collected as time-series data by the calculation means, and this time-series data, etc. Based on this, create a heater temperature target value pattern during cooling operation. This heater temperature target value pattern may be created, for example, as a table of time series data, or as a function of time.

Next, during the main cooling operation with the wafer inserted into the furnace,
The created heater temperature target value pattern is reproduced and used as the heater temperature target value, and the heater temperature is controlled using the output of the heater temperature detector as a feedback signal.

Therefore, even during single cooling operation, the furnace temperature is accurately controlled to the furnace temperature target value without being affected by heater temperature detection errors. In addition, if the semiconductor processing furnace is composed of multiple zones, the heater temperature target value pattern described above is created for each zone, and during the main cooling operation, each heater temperature target value pattern is used to create the heater temperature target value pattern for each zone. By controlling the heater temperature using the detected temperature value of each zone heater as a feedback signal, the temperature difference in the furnace between the zones can be minimized.

(Example) Hereinafter, an example of the present invention will be described based on the drawings. This embodiment is applied to a temperature control device for a semiconductor diffusion furnace.

1 to 4 are diagrams showing one embodiment of the present invention.

First, to explain the configuration of the temperature control device for a semiconductor diffusion furnace, FIG. 1 shows the heater, temperature control device, etc. for one zone in the semiconductor diffusion furnace.
5 is a heater almost similar to that shown in FIG. 6; 26;
is a heater temperature detector attached to this heater 25;
7 is the heater power supply, 28 is inside the quartz tube (inside the furnace) when measuring the furnace temperature.
The heater 25 is heated by the power supplied from the heater power source 27, the heater temperature changes, and the furnace temperature is further increased through the furnace process 29. There is. In addition, although not shown in FIG. 1, the semiconductor diffusion furnace includes the t shown in FIG.
A blower is installed as a cooling means similar to a blower.

Reference numeral 30 denotes a temperature control device that controls the temperature inside the furnace, and inside thereof is a first calculation device 31 that collects and analyzes actual data of heater temperature detection values during forced air cooling in advance.
, a second arithmetic unit 32 that creates a heater temperature target value pattern during forced air cooling operation based on this performance data, and a furnace temperature detection value θF( 'C) as a feedback signal, the first PID calculation device 33 serves as a first temperature control device for controlling the furnace temperature to the target value θFR ('C), and the heater temperature during the main cooling operation with the wafer inserted. A second control for controlling the heater temperature detection value θN ("C), which is the output of the detector 26, as a feedback signal to the heater temperature target value θHR (°C) created by the second arithmetic unit 32. Second PID* frame device 34 as means, heater power supply 27
(7) Heater power supply driver 35 that manages ONloFF
etc. are provided.

The first performance device 31 has a detected heater temperature value θH (
'C) and the furnace temperature detection value θF ('C) are taken in, and when the forced air cooling start timing signal 36 is given when the wafer is not inserted, the heater temperature detection value Actual data is collected as time series data, etc. θFB('C) is the target value of the furnace temperature after forced air cooling, T(
sec) is the time from the start of forced air cooling to the start of data sampling, and E ('C) is a parameter for the data sampling stop condition. +E) (”C
) Data sampling of actual data is stopped when the following conditions are reached. This first arithmetic device 31
Then, C is sent to the second arithmetic unit 32 as a parameter to be described later that indicates a heater temperature target value pattern during forced air cooling.

In the second arithmetic unit 32, when the forced air cooling start timing signal 3B is given after the wafer is inserted, a heater temperature target value pattern θHR during forced air cooling is generated according to the furnace temperature control sampling timing signal 39, and this is the second
is set as the target value of the PID calculation device 34.

The above-mentioned first and second computing devices M31.32 collect the actual data of the heater temperature detection value while the first PID computing device 33 is controlling the temperature inside the furnace, and use this actual data. A calculation means is configured to create a heater temperature target value during cooling operation from the above.

41.42 is the control output of the first and second PID calculation devices 33.34, which indicates the ON rate of the heater power supply 27, that is, what percentage of the heater power supply 27 is turned on during a certain cycle time.
This represents a command indicating whether to turn on the switch. 43.4
Reference numeral 4 denotes a switch for switching the control method, and the switch 43 is closed in the preliminary operation when the wafer is not inserted, and the switch 44 is closed in the main operation when the wafer is inserted. The switches 43 and 44 are interlocked so that when one closes, the other always opens. 45 is finally heater electric 1! i
27 is the power supply 0N10FF signal.

Next, the operation of the semiconductor diffusion temperature control device configured as described above will be explained with reference to FIGS. 2 to 4.

FIG. 2 is a flowchart showing the processing contents in the first arithmetic unit 31, and (A>, (B) in FIG. 3) shows that the switch 43 in FIG. 1 is closed when no wafer is inserted, and the detected value of the furnace temperature is fed back. As a signal, it shows the behavior of the detected heater temperature value θt1 and the detected furnace temperature value θF when strong f and IJ air cooling operation are performed, and Fig. 4 shows the cooling pattern of the heater temperature target value. be.

As shown in FIG. 3, from the time when forced air cooling was started and cooling air was drawn into the furnace by the blower, the detected heater temperature value θ■ rapidly decreased. This shows the influence of the heater temperature detector 26 being agitated by the cooling air mentioned above. Then, after approximately Tsec has elapsed after the forced air cooling was started, the heater temperature θ■ decreases almost linearly. Therefore, the heater temperature detection value during forced air cooling can be approximately approximated by the cooling pattern shown in FIG.

The first arithmetic unit 31 has the main function of collecting and analyzing the performance data during the above-mentioned forced air cooling operation to obtain the parameters C and K shown in FIGS. 3 and 4. Hereinafter, the processing contents of the first arithmetic unit 31 will be explained using the flowchart shown in FIG.

After f1 forced air cooling is started, a parameter j representing the number of data sampling times is set to zero in step 51. Next, in step 52, the elapsed time t after the start of forced air cooling and (T
+j”TS).Here, King is the time (s) from the start of forced air cooling until the start of data sampling.
ec), and the time required for the heater temperature detection value θH in FIG. 3(A) to start falling linearly after the start of forced air cooling (usually about 250 sec) can be given. Ts is the data sampling pitch. Since t<(T+J・Ts) until the data sampling timing is reached, the timer counts up in step 53, and when the data sampling timing is reached, t≧(T
+j-TS), so the process advances to step 54.

Step 54 indicates the end condition of data sampling, and the detected furnace temperature value θF is equal to the target furnace temperature value θFB after forced air cooling.
Conditions are set so that data sampling is terminated when the distance is sufficiently close to . The parameter for this is E (“C), which is the difference between θF and θFB (θF−θF
Compare l+)(''C) and E('C), (θF-θF
B) If ≧E, in step 51 tj(s e c
) and θHj('C). Here, tj and θHj are the elapsed time l! of the data sampling time as shown in FIG. 3(A). It (sea) and detected heater temperature value θH ('C). Furthermore, step 5
Step 6 counts up the parameter j representing the number of data samplings (j-J+1), and then returns to step 5.
Return to 2.

In this way, while (θF−θF11)≧E, T
s (sec), tJ and θHj are memorized, and time series data of the heater temperature detection value θH ("C) is obtained. Furthermore, forced air cooling operation progresses, and the furnace temperature detection value θ
Data sampling ends when F approaches the target value θFB of the furnace temperature after forced air cooling and (θF−0FB)<E in step 54.

Assuming that the value of E ('C) is approximately 30 ('C), it is possible to sample the almost linear portion of the time-series change in θH ('C), as shown in Figure 3 (A). .

After data sampling is completed, the processing from step 57 onwards is performed. n in step 57 is a parameter necessary for the process in step 58, and indicates the total number of data sampling times. In step 58, the stored time series data (1J) is stored.

θH, ; (j-0, 1, 2, . . . , n-1>) is linearly approximated, and the parameters K and C when approximated by the cooling pattern shown in FIG. 4 are calculated. The values of K and C are time series data ('jjsθoj (j-0, 1, 2, ...,
n-1)) is calculated by the following formula by first-order approximation using the method of least squares.

K-(n-W-X-Y)/(n-A-X2)('C/5
ec)...(1) C-θH^-(A-Y-W-X) / (n-A-X2) ("C)-" (
2) However, A-Σ(tj)2...(4)
sQ 1%-1. The process of deriving the above equations (1) to (5) will be described in detail in other embodiments to be described later.

That's the first computing device! This is the processing content of No. 31.

Since the cooling pattern of the heater temperature detection value θH during forced air cooling has been obtained by the first arithmetic unit 31, the switch 44 is closed during the forced air cooling operation during wafer insertion, and control is performed using the heater temperature detection value θH as a feedback signal. Switch to . The heater temperature target value at this time is output from the second arithmetic unit 32 according to the cooling pattern shown in FIG. 4 using 1G as the above-mentioned parameter, and the heater temperature is controlled. The heater temperature target value pattern θHR output from the second arithmetic unit 32 is the detected heater temperature value θI during preliminary operation.
(Since the cooling pattern of ('C) is simulated, this θ
By controlling the heater temperature with the second PID calculation device 34 using on('C) as the heater temperature target value,
Furnace temperature detection value θF (
'C) behavior is almost reproduced.

Note that for convenience of explanation, Figures 1 and 2 focus on one of the multiple zones and show its configuration and processing content, but the configuration and processing content will be similar for other zones. The explanation will be omitted here.

As described above, the temperature control device for a semiconductor processing furnace of this embodiment directly detects the temperature inside the furnace during preliminary operation, accurately controls it to the furnace temperature target value, and adjusts the detected heater temperature value at this time. By collecting and analyzing operational data of
Even during the actual operation, when the furnace temperature cannot be directly detected, the furnace is configured to reproduce the same temperature as during the preliminary operation, so the furnace temperature that occurs due to errors in heater temperature detection during forced air cooling can be reduced. Control errors and furnace temperature differences between zones are eliminated, and highly accurate furnace temperature control can be achieved during forced air cooling operation.

In addition, since the data collected during preliminary operation also reflects the influence of mechanical constants such as the characteristics of the heater and the passage characteristics of cooling air, the control system should be adjusted to take into account the influence of mechanical constants that differ depending on each diffusion furnace. becomes almost unnecessary.

Next, another embodiment will be described using the flowchart of FIG.

In the above embodiment, a cooling pattern that approximates the heater temperature detection value during the preliminary operation is used as the heater wetness target value during the main operation. For this reason, if the delay in the heater temperature control system is large, the reproducibility of furnace temperature changes during preliminary operation may be reduced. In that case, after the output of the second arithmetic unit 32 in FIG.
・Compensation elements such as s) (where S is the Laplace operator,
T1 and T2 are constants) can be inserted to take measures against the above-mentioned delay.

In addition, in the flowchart of FIG. 2 showing the processing contents of the first arithmetic unit 31 in FIG. 1, the time series data (tJN θ
hj (j=o, 1.2, . . . , n-1)), but in order to save memory, processing as shown in the flowchart of FIG. 5 may be performed.

Here, the basic concept of memory saving seen in the flowchart of FIG. 5 will be explained below.

The time series data (1,;1
As shown in Figure 4, θHj (J-0, 1, 2, ..., (n-1)) is expressed by the following formula, θo; ，
2, ..., (n-1> ...(6) Consider the case of first-order approximation (K, C: constants to be determined by least squares method).

In equation (6), if the approximation error is ε, then θHo=−To+(θHA−C)+ε0θH1=−
t1+ (θH^-C)+ε1θ811-1= K 1
1tn-++ (θHA-C)+ε-...(7) The quadratic evaluation function for the approximation error in equation (7) is J=Σε
・2...(8) Consider setting j・0″l and minimizing J.

-(θHA-C))2 (9) Therefore, the necessary and sufficient conditions for J to be minimum are the following equations (10) and (11).

-(θH^-C)) ・(-tj)=0...
(10) −(θHA−C))=0... (11)(10), (
11) From formula h−啜... (13) Furthermore, by expanding this, we get -Σt・・θHj...(14)
-OJ bar 1 - Σ゛θHj ... (15) IIO If we summarize this in matrix form, ... (17) W; Σtj ・θnj ... (
18) aQ.

From equation (16), KlG is calculated as K-(n-W-X-Y)/(n-A-X2)...
(19) C=θHA-(A-Y-W-X)/(n-A-X2) (20).

In this case, the value of C is based on time series data (t・, θHj (j=
After the collection of 0.1.2,..., n-1)) is completed,
Since this involves calculation, if there is usually enough memory, a method may be adopted in which the time-series data is temporarily stored. This is the method shown in Figure 2.

However, if there is not enough memory, as shown in steps 61 to 68 in FIG. 5, the parameters X and Y1A necessary for calculating K and C.

Update W sequentially in parallel with data sampling,
As a result, a method can be adopted in which equations (17) and (18) are calculated.

The differences between the processing content shown in FIG. 5 and the processing content shown in FIG. 2 are as follows.

In the process shown in FIG. 2, the detected heater temperature value θ14 J (°C) during data sampling and the elapsed time tj (S e G ) after the start of forced air cooling are converted into time series data (1j).

C
is being calculated.

On the other hand, in the process shown in FIG. 5, during data sampling, time-series data storage is not performed as in the process shown in FIG.
) and (5), memory is saved by adopting a method of sequentially updating the values of ×1Y, A, and W at each data sampling timing.

In addition, in the processing shown in FIGS. 2 and 5, the cooling pattern of the heater temperature detection value during forced air cooling is approximated to the first order, but it is of course possible to approximate this by a higher-order curve of second or higher order. , time series data (t, ,
The pattern of the heater temperature target value θHR may be created by interpolating and extrapolating θHj (j=0.1, 2, . . . , n-1).

[Effects of the Invention] As described above, according to the present invention, when setting the furnace temperature to a desired furnace temperature target value, the furnace temperature detection value is directly used as a feedback signal to set the furnace temperature II. ! A pre-cooling operation is performed to control the furnace temperature to the A value, and during this pre-cooling operation, actual data of the heater temperature detection value is collected, and based on this actual data, the heater temperature target during the cooling operation is determined. A value pattern is created, and during the main cooling operation, this heater temperature target value pattern is used as the heater temperature target value, and the heater temperature detection value is used as a feedback signal to control the heater temperature. Therefore, during the main cooling operation, the heater temperature detection error is It is possible to accurately control the temperature inside the furnace to Ftliil without being influenced by the temperature of the furnace.

In addition, when a semiconductor processing furnace is composed of multiple zones,
The above heater temperature target value pattern is created for each zone, and during the main cooling operation, each heater temperature target value pattern is set as the P91 degree target value for each zone, and each zone heater temperature detection value is used as a feedback signal to control the heater temperature. By doing so, the temperature difference in the furnace between each zone can be suppressed to a minimum.

[Brief explanation of the drawing]

1 to 4 show an embodiment of the temperature υ11II apparatus for semiconductor processing furnaces according to the present invention, FIG. 1 is a block diagram, and FIG. 2 shows processing contents of the first arithmetic unit section. Flowchart, FIG. 3 is a characteristic diagram showing the behavior of the heater temperature detection value and furnace temperature detection value during pre-cooling operation, FIG. 4 is a characteristic diagram showing the heater temperature target value pattern, and FIG. FIG. 6 is a flowchart showing the processing contents of the first arithmetic unit portion on the flow side, and FIG. 6 is a longitudinal cross-sectional view showing the furnace structure of a conventional semiconductor diffusion furnace. 23 Niblower (cooling means), 25: Heater, 26: Heater temperature detector, 28: Furnace temperature detector, 31: First computing device, 32: Second constituting computing device together with the first computing device.
arithmetic device; 33: first PAD arithmetic device (first temperature control means); 34: second PID arithmetic device (second temperature control means).

Claims (1)

[Claims] A temperature control device for controlling the temperature inside a semiconductor processing furnace, which is equipped with a heater for raising the temperature inside the furnace and a cooling means for lowering the temperature, the temperature control device being used in advance to set the temperature inside the furnace to be lowered to a required furnace temperature target value. a first temperature control means for controlling the furnace temperature to the target furnace temperature value using the detected value of the furnace temperature as a feedback signal during cooling operation; a calculation means for collecting actual data of heater temperature detection values of the section and creating a heater temperature target value pattern during cooling operation from this actual data; and a heater temperature target value pattern created by the calculation means as a heater temperature target value. 1. A temperature control device for a semiconductor processing furnace, comprising: a second temperature control means for controlling the heater temperature during cooling operation using a detected heater temperature value as a feedback signal.