JP2818689B2 - Development method - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of developing a register in a photolithography process included in a process of manufacturing a semiconductor integrated circuit device.

Conventional technology Conventionally, as a method for keeping the finished dimensions of a resist pattern after development constant, light is irradiated from the top surface of the resist to the substrate, the change in reflected light intensity is measured, and the development time is adjusted based on the result. The method of doing was known.

 The outline of the method is as follows.

As time elapses from the start of development, a resist pattern such as BCDE is formed on the resist 37 on the base 36 as shown in FIG. At this time, resist
When light is applied to the base 36 from above 37 and the reflected light is detected, the change in the intensity of the detection signal becomes a waveform having peaks and valleys as shown in FIG. This is because interference occurs between light reflected from the surface of the resist 37 and light reflected from the surface of the base 36, and a waveform having peaks and valleys appears. In FIG. 2, A, B, C... Correspond to the patterns A, B, C. At the point A, the development has not yet been performed, but peaks and valleys of the waveform occur as the development proceeds. The cycle of the peaks and valleys corresponds to the fact that the resist film thickness is developed by λ / 2n. λ is the wavelength of the light being monitored, and n is the refractive index of the resist. At the time point C, the resist remains by λ / 4n, and at the time point D, the development has progressed to the base and a part of the base has been exposed. No peaks and valleys appear after point D, and the reflected light intensity gradually increases as the exposed area of the base increases.

In the conventional method, the time until the last minimum point C or breakthrough point D appears is detected, and the total development time is determined as a function of the time. However, since the developing device and the reflected light intensity measuring device are generally different from each other, an interface is required between the two devices in order to adopt this method, which may be difficult.

There is also a problem in the inspection of the resist pattern dimensions after development. According to the conventional inspection method, length measurement by an optical dimension measuring device and scanning electron microscope (length measurement SE)
M) is 1 to 4 per 50 product wafers
It was sampled and inspected at the rate of one sheet. However, the length measurement by an optical device has a problem that it is difficult to measure a pattern of 2.0 μm or less because of a large error. The latter has a high length measurement accuracy, but is measured in a vacuum.
There is a problem that it takes about 10 minutes to measure the points and the throughput is low.

In addition, contamination of the measured wafer is also a problem. That is, in the case of the optical method, there is a problem of adhesion of dust due to the transfer of the wafer, and in the method of the length measurement SEM, since the electron beam is irradiated in a vacuum, various gases are generated and the wafer surface may be contaminated. It is a problem.

In addition, although product inspection is ideally ideal for product inspection, it is impossible to inspect all products because of the low throughput of either inspection method, and it is not possible to eliminate the risk of defects existing in uninspected wafers There was also a problem.

In addition, the photoresist is sensitive (Et) due to variations in substrate reflectivity, resist film thickness, pre-bake temperature, time, etc.
h) changes. Therefore, 1 μm L / S in 60 seconds development
(Line / space) = 1: 1 exposure amount is 100 (mJ / c, for example)
m ² ), the same line width is not always obtained on the next day under the same conditions, even if the exposure is 100 (mJ / cm ² ). There was also a problem.

SUMMARY OF THE INVENTION The present invention provides a new developing method which substantially eliminates the need for the conventional inspection having the above-mentioned problems.

Means for Solving the Problems The present inventors have paid attention to the fact that the development end point fluctuates by linking to the aforementioned process parameters such as the reflectance of the substrate and the resist film thickness. The present invention has been completed by finding that the exposure amount necessary to obtain a predetermined line width can be derived from the development end point when development is performed with a known exposure amount.

In the present invention, the development end point is determined by the following method.

In the present invention, either the last minimum point or the breakthrough point may be used as the development end point.

(1) Method for detecting the last minimum point The method for detecting the last minimum point in the development process is as follows: The method for detecting the time until the minimum point C appears is as follows.
A template method has been proposed. (US Patent 4,
647,172 `` Resist Development Mathod '' GCA Corpora
This is achieved by irradiating the substrate with light from the upper surface of the resist, detecting the reflected light, and superimposing the data of the intensity change actually obtained on the data (template) stored by the computer to obtain the minimum point C. Is a method for detecting

However, in this method, the minimum point C cannot be detected unless the number of data constituting one fringe is the same.

If the number of data constituting one fringe is different when the minimum point C is detected, the following algorithm may be used.

Basic end-fringe algorithm Applied when the cyclic elongation rate does not change much during the development process in a sine wave development waveform.

End-fringe algorithm flowchart 3
Shown in the figure.

As parameters, an algorithm operation start time α from the measurement start point of the computer and an allowable period elongation rate β (> 1.0) are input. The algorithm operates after exceeding α after the start of development. If the next minimum point appears before the time interval Xβ of the last two minimum points known at the time t, the calculation is further executed. If the next minimum point does not appear until the time interval Xβ between the last minimum point that is known at the time t, the last minimum point is detected as CPT.

Algorithm End Time Designation Type End Fringe Algorithm This algorithm calculates the cycle elongation β of the obtained development waveform.
If is not extremely large, the CPT cannot be detected, and therefore the present invention is applied to a case where the CPT cannot be detected only after the total development time TDT. FIG. 4 shows a flowchart of the present algorithm. The algorithm end time γ is input as a parameter. After the start of development, the last minimum point known at time γ is defined as CPT.

Algorithm End Time Designation Type Developer Discharge Time Adjustable End Fringe Algorithm This algorithm is applied in the process of developing a photoresist whose time to the minimum point C is extremely short.

FIG. 5 shows a flowchart of the present algorithm. As parameters, an algorithm end time γ, a standard developer discharge time E, and a standard time CPTe up to a minimum point C are input.

When the algorithm end time comes, the last minimum point C known at that time is detected as CPT. At this time, the discharge time of the developer is S: developer discharge time E: standard developer discharge time CPT: time to minimum point C CPTe: time to standard minimum point C Calculated by the above equation.

This algorithm is applied to a developed waveform where one large valley appears.

The input parameters are the algorithm start time α, and the maximum waiting time Tw or ratio multiple when waiting for the next minimum point to appear after one minimum point is found.

The last minimum point C detected before Tw after elapse of
Detected as T.

 FIG. 6 shows a flowchart of the present algorithm.

Minimal point search type back search algorithm This shows a sine curve shaped development waveform, and is applicable when large noise occurs in the waveform after CPT.

FIG. 7 shows a flowchart of the present algorithm. The input parameters are the time Bα during which the back search algorithm is to operate and the threshold voltage Vth.

From time Bα, the waveform is examined in the direction opposite to the time axis, and a portion exceeding the voltage Vα ± Vth at time Bα is searched. From there, the thorn shape is further examined in the direction opposite to the time axis, and the minimum point C is detected.

It works in the same way as the local maximum search back search algorithm. This algorithm is used to detect the last local maximum if the last local minimum C is ambiguous.

 FIG. 8 shows a flowchart of this algorithm.

Estimated curvature detection algorithm This algorithm detects the penultimate minimum as point C if the shape of the fringe containing the last minimum is significantly different from the shape of the fringe containing the penultimate minimum. Sometimes apply.

As an input parameter, the estimated time Es is input. At the time of Es, the curvatures of the right and left fringes are compared with respect to the fringe including the voltage data at the time of Es. At this time, the minimum point of the immediately preceding fringe from the minimum point of the fringe having a different curvature is detected as CPT.

 FIG. 9 shows a flowchart of this algorithm.

(2) Method of detecting breakthrough point The above algorithm is for detecting the last minimum point when the CPT is used. However, the breakthrough point may be used as the CPT. An algorithm for detecting this is as follows, for example.

No Fringe Algorithm This algorithm is applied to a developed waveform having one inflection point.

In order to detect an inflection point, a second derivative with respect to time of the data (volt) is obtained, and a point at which the second derivative reaches a maximum is detected as a CPT.

 FIG. 10 shows a flowchart of the present algorithm.

These algorithms can be used alone or in combination.

Further, in the above-described developing method, it is preferable that a stop is provided in the irradiation light source or the light receiving unit, and the stop is adjusted for each wafer so that the received light intensity of the reflected light is appropriate for CPT measurement. This adjustment is performed before the resist is applied to the wafer. The wafer is irradiated with light before the application of the resist, the reflected light intensity is measured, and the aperture is adjusted so as to obtain a predetermined reflected light intensity.

FIG. 11 shows a change in the intensity of the reflected light when the reticle is developed with an exposure area of about 25% using an irradiation light amount of 3V. In this case, the irradiation light amount is within an appropriate range, and the change in the intensity of the measured reflected light is clear. FIG. 12 and FIG. 13 show changes in the intensity of the interference liquid when the irradiation light amount is 0.5 V and 8 V. It can be seen that the intensity change cannot be accurately measured because the irradiation light quantity is too small or too large. Eleventh
If a clear change in intensity can be measured as shown in the figure, the CPT
Can be measured accurately, and as a result, the finished dimension LW
Can also be accurately determined.

The adjustment of the aperture may in principle be performed manually.
However, during the semiconductor manufacturing process in the production line, since a film having various reflectances is formed in multiple layers on each semiconductor wafer by each process, when measuring the intensity change of the above-mentioned interference wave, the photoresist layer is required. If the lower layer is a film with low reflectance, the aperture of the light source for reflected light is set for low reflectance. Conversely, for a film with high reflectance, the aperture of the light source for reflected light is highly reflected. It is troublesome to manually adjust the aperture of the light source each time the type of the formed film is different for the rate.

In recent years, in the field of semiconductor integrated circuit manufacturing, the minimum line width of a circuit pattern handled in a photolithography process has become increasingly smaller and has reached a so-called submicron region of 1 micron or less. Also, miniaturization technology has been required for each apparatus for development.

In order to obtain such a submicron line width, it is necessary to measure the reflectivity of each film of each wafer even if the film type is the same (for example, SiO ₂ ), and to always grasp subtle changes in the reflectivity. Is required.

Not only is it necessary to manually adjust the aperture of the light source and set the amount of light appropriately when the type of film is known (for example, from SiO ₂ to Al), but also for each wafer, productivity has to be improved. There is a problem above.

Further, it is important to suppress the generation of dust, dirt, and the like in a semiconductor manufacturing line, and it is therefore desired to eliminate as much as possible manual work.

In consideration of the above points, it is preferable to automatically adjust the aperture.

 FIG. 14 is a schematic diagram of the developing device.

Reference numeral 21 denotes a chuck of a spin device, which is rotated by a rotating shaft 22. 33 is a motor drive circuit for rotating the motor of the spin device. The chuck 21 is provided with a suction mechanism, and the base 23 is attached by suction. Above the base 23, a developer supply mechanism 24 and a rinsing liquid supply mechanism 25 are provided.
And are provided. A developing solution is sprayed from the developing solution supply mechanism 24 in a band shape having a width of about 1 cm and a length of about 10 cm in a radial direction of the base 23, and the base 23 is rotated by a spin device so that the developing solution is uniformly spread on the base 23. Is poured. The rinsing liquid supply mechanism 25 can similarly spray the rinsing liquid onto the base 23.

Reference numeral 26 denotes a light source having a long wavelength, which is an optical system for measuring the film thickness of the resist on the base 23. For example, a tungsten lamp or a mercury lamp is used. Reference numeral 27 denotes a filter for transmitting light having a wavelength at which the resist is not exposed among the light from the light source 26. For example, a filter for blocking light having a short wavelength of 460 nm or less is used. Reference numeral 28 denotes an optical fiber bundle for guiding the light passing through the filter 27 onto the base 23, and irradiates the light so that the diameter of the light on the base 23 is, for example, about 10 mm.

The optical fiber bundle 28 also serves to guide light reflected from the substrate 23. The reflected light guided by the optical fiber 28 is received by the phototransistor 30 via the filter 31. The filter 31 is a narrow band-pass filter that transmits only light in a specific wavelength range.
Use one that can select a wavelength that increases the S / N ratio. The detection signal of the phototransistor 30 is converted into a digital signal through an A / D converter 35, taken into a computer 32, and analyzed. As the computer 32, a commercially available personal computer can be used.

FIG. 15 shows an example of a device for automatically adjusting the aperture. When the semiconductor wafer 41 is placed on the check 42 of the developing device, the intensity of light reflected from the film 45 on the semiconductor wafer 41 is measured before applying the photoresist liquid. The comparator 46 compares the difference between the measured value and the appropriate amount of light determined to be applied to the film developed before the film 45. When the difference is within the allowable error, the process proceeds without performing the adjustment.
If the difference is larger than the allowable error, an instruction is output from the comparator 46 to the adjustment system 47 so that the aperture driving motor 44
And adjust so that the difference is within the tolerance.

(3) Method of polynomial approximation and interpolation To formulate the relationship between LW and CPT, it is sufficient to express LW as a CPT polynomial by analyzing LW and CPT data obtained by experiments. The value of LW corresponding to an arbitrary CPT can be obtained by interpolation.

Polynomial approximation and interpolation are performed by any of the following methods, or by a combination of the following methods.

Polynomial interpolation The relationship between LW and CPT can be formalized by determining each coefficient of ko to kn using the n-th order polynomial of the above equation. It goes without saying that increasing the order improves the accuracy of the approximation,
Even when an accurate approximation is required, it is sufficient that the order is at most ten, and in the ordinary case, the fifth or sixth order expression is sufficient.

Further, the effects of the present invention can be sufficiently obtained even in the case of the second-order or third-order formula. The determination of the order of the approximate expression is determined only by the management width allowed in product management.

Depending on the form of the interpolation formula by fitting to the linear regression formula by variable conversion, it may be possible to express a linear relationship by performing appropriate variable conversion.

 Well-known examples are listed below.

For example, if y = ae ^bx , take the logarithm of both sides
1ny = 1na + bx, and when this is compared with Y = A + BX, the relationship described in the above table is obtained.

 Interpolation by Lagrange supplementary equation The Lagrange supplementary equation refers to the following equation.

This equation is a polynomial composed of (n + 1) terms of order n with respect to x. When using the Lagrange interpolation formula, it is necessary to pay attention to the accuracy of the entire interpolation formula is reduced if even one set includes data having low reliability.

Interpolation by B-spline interpolation formula The theory of B-spline was first shown by Schoenberg ¹⁾ , and Cox ²⁾ and de Boor ³⁾ found an iterative formula that was effective for numerical calculation of splines. Riesenfeld ⁴⁾ applied the B-spline basis function to the definition of the curve.

The B-spline curve p (t) represented by the function of the parameter t is represented by the following equation.

Here, Pi represents a position vector of the curve definition polygon.

Order k, the i-th normalized B-spline basis function Ni,
k (t) is given by the following iterative formula.

x _i is an element of the knot vector.

For details on B-spline interpolation, see "Computer Graphics" (translated by Fujio Yamaguchi, May 1979)
30th, published by Nikkan Kogyo Shimbun).

1) Schoenberg, IJ: “Contributions to the Probl
em of Approximation of Equidistant Data by Analyti
c Functions "Q.Appl.Math., Vol.4 (1946) p.45-99,112
~ 141.

2) Cox, MG: “Tho Numerical Evaluation of B-Sp
lines "National Physical Laboratory DNAC4, August (1
971).

3) de Boor, Carl: “On Calculating with B−Spline
s "J. Approx, Theory, Vol. 6 (1972) pp. 50-62.

4) Riesenfeld, RF ,: “Berstein-Bezier Methods f
or the Computer-Aided Design of Free-From Curves
and Surfaces ", Ph, D.Thesis, Syracse University, Marc
h (1973).

By any or a combination of the above methods, any
LW for CPT can be determined.

Similarly, the exposure amount for an arbitrary CPT can be determined.

The present inventors have experimentally found that the relationship between the development end point and the exposure amount can be expressed by an exponential function. That is, the relationship between the two can be represented by the following general formula.

E = kxCPT ⁿ ... 1) Here, E indicates the exposure amount, CPT indicates the time until the end of development, and k and n are constants. Note that the CPT may be a time until the last minimum point or a time until a breakthrough point, as described above.

Further, the relationship between CPT and line width can be expressed by a polynomial.

That is, the relationship between the two can be represented by the following general formula.

LW = k ₀ = Σki (CPT) ⁱ ... 2) where LW is a finished dimension, k ₀ is a constant, and k _i is a coefficient of the i-th term. The finished size refers to a value obtained by measuring a photoresist size after completion of development by a fine line width size measuring device (commonly called a length measuring SEM) using a scanning electron microscope. The definition of the dimension refers to the width of the bottom of the resist pattern (the width of the resist line at the portion bonded to the substrate).

Conversely, CPT can be expressed as an LW polynomial as follows.

CPT = a ₀ + Σai (LW) ⁱ ... 3) Therefore, according to the above equations 1), 2) and 3), CPT,
The relationship between E and LW can be expressed, and it can be seen that LW can be kept constant by adjusting E according to CPT.

 Specifically, the exposure amount can be determined by the following method.

1) Measure the CPT at the exposure amount Ep as CPTp.

Pseudo exposure E _Q when the 2) CPTp obtained by equation 1.

3) The CPT that gives the target line width LWn is obtained by Expression 2.
(CPTn).

4) The exposure amount En in CPTn is obtained by Expression 1.

5) Assuming that the exposure amount is Ep, the correction exposure amount E _R and the exposure correction amount ΔE _R for obtaining the target line width LWn are as follows.

When CPTn <CPTp logE _R = logEp + (logEn−logE _Q ) logΔE _R = logEn−logE _{Q When} CPTn> CPTp logE _R = logEp + (logE _Q −logEn) logΔE _R = logE _Q −logEn E when CPTn = CPTp _R = Ep ΔE _R = 0 If exposure is performed with the exposure amount E _R obtained by the above method,
CPT becomes a predetermined value, and therefore, a target line width LWn can be obtained. When the exposure amount is changed according to the CPT as described above, there is an advantage that the development time can be kept constant. That is, since the exposure amount is adjusted so that the CPT becomes a constant value, the development time is also constant as a result. Therefore, the developing device does not need to change the set time during the operation, and the development becomes easier and more stable.

According to the present invention, the exposure time of the next wafer is corrected using the past development end point value instead of performing the exposure correction in real time. Therefore, for example, the following applications can be considered.

1) Detection of end point of preceding inspection wafer and exposure amount correction system for subsequent wafers For example, the first wafer of a lot is developed, and the development end point is detected. Whether the wafer is within the standard is determined by Expression 1, and the required exposure amount for obtaining a predetermined line width is calculated from Expressions 2 and 3. Under the assumption that the process conditions in one lot do not change much, the required exposure amount obtained earlier is sent to the exposure machine, and the exposure amount of the subsequent wafer is substituted with the required exposure amount.

This system enables the advance inspection process to be automated. Further, it goes without saying that dimensional control can also be performed on the wafer processed with the required exposure amount using the equation (1).

2) A system that analyzes seasonal process fluctuations by monitoring the exposure correction amount over the long term. Monitors the required exposure amount of wafers that are processed in large quantities on the production line, and collects data for each lot, day, and month. Can be used as a database for a system that analyzes seasonal process fluctuations.

The line width of the photoresist obtained by the developing method according to the present invention shows a very good agreement with the expected line width. As shown in the examples, when the line width is controlled by the developing method of the present invention, more preferable results are obtained than when the process is controlled by optical length measurement.

Further, since it is not necessary to perform length measurement as a practical operation as in the conventional method, there is no need for an operator for length measurement, there is no reduction in throughput, and virtually all product inspection is performed. There is an advantage that the same effect can be expected.

Example 1 TSMR-V3 manufactured by Tokyo Ohka Kogyo Co., Ltd. was used as a photoresist, prebaked at 80 ° C. for 60 seconds, and then baked at 100 ° C. for 60 seconds. As a developer, a 2.38% aqueous solution of tetramethylammonium hydroxide was used.

The resist film thickness was 1.011 μm, and the exposure was 144,153,1.
87, and 221 (mJ / cm ² ), and the development end point and finished dimensions were measured.

Next, with the exposure amount set to 170 (mJ / cm ² ), the resist film thickness was changed to 0.9649, 0.9829, 1.0089, 1.0144, and 1.0385 μm, and the end point of development and the finished dimensions were measured.

Table 1 shows the measurement results. FIG. 16 shows a graph of the development end point and the finished dimension. From these results, the relationship between the end point of development and the finished dimensions was approximated as the following fourth-order function polynomial, and each coefficient was determined. The following results were obtained.

_{LW = k 0 + k 1 xCPT} + k 2 xCPT 2 + k 3 xCPT 3 + k 4 xCPT 4 k 0 = -1.096 k 1 = 0.7356 k 2 = -0.09902 k 3 = 0.006081 k 4 = -0.0001403 However, the scope of the CPT 5.8 ≦ CPT ≦ 13.6.

The relationship between CPT and exposure was approximated as the following exponential function, and the coefficient was determined. The following results were obtained.

E = kxCPT ⁿ k = 6.097 × 10 ² n = −0.5856 FIG. 17 shows the relationship between CPT and exposure.

FIG. 18 shows the correspondence between the expected dimensions obtained by the developing method according to the present invention and the dimensions actually measured by SEM. The correlation coefficient is 0.
99293, indicating a very good correlation.

Example 2 A resist film having a thickness of 1.011 μm and an exposure amount of 170 and 204 (mJ /
cm ² ), the exposure amount is 170 (mJ / cm ² ), and the resist film thickness is
The development was performed under the conditions of 0.9829 and 1.0222 μm, and the other conditions were the same as in Example 1.

Table 2 shows the line width predicted based on the approximate expression obtained in Example 1 and the line width actually measured by the SEM. Both show very good agreement.

SEM measurement results obtained in Example 1 and Example 2,
FIG. 19 shows the results of each optical measurement. The results obtained by optical measurement do not give satisfactory results, and it can be seen that the developing method according to the present invention gives more preferable results than the optical measurement.

As can be seen from the above examples, the developing method according to the present invention makes it possible to accurately predict the finished line width from the CPT, and changes the CPT by changing the exposure based on the relationship between the exposure and the CPT. It can be seen that the desired finished size can be obtained by adjusting the. Furthermore, the value agrees very well with the measurement result by the length measurement SEM, and gives a better result than the optical measurement.

[Brief description of the drawings]

FIG. 1 is a cross-sectional view of a resist showing a development process with time, FIG. 2 is a waveform showing an example of a change in reflected light intensity, FIG. 3 is a flowchart of a basic type end-flind algorithm, and FIG. FIG. 5 is a flowchart of an algorithm end time designating type developer discharge time adjusting type end flind algorithm, FIG. 6 is a flowchart of an Ichitani fringe-compatible end flind algorithm, FIG. Flowchart of the point search type back search algorithm, FIG. 8 is a flowchart of the maximum point search type back search algorithm, FIG. 9 is a flowchart of the estimated curvature detection type algorithm, FIG. 10 is a flow chart of the no fringe algorithm, FIG. Fig. 13 shows the irradiation light A graph showing the change in the intensity of the interference wave when developing using a reticle with an exposure area of about 25% at an amount of 3 V, 0.5 V, and 8 V, respectively. FIG. 14 is a schematic diagram of the developing device, and FIG. Figures 16 to 19 are graphs showing the results of the examples. EXPLANATION OF SYMBOLS 21: Spin device chuck 22: Rotating shaft, 23: Substrate 24: Developer supply mechanism 25: Rinse solution supply mechanism, 26: Light source 27, 31 Filter 28: Optical fiber Bundle 30 Phototransistor 32 Computer 33 Motor drive circuit 35 A / D converter, 36 Underlayer 37 Resist 41 Semiconductor wafer 42 Chuck 43 Light source 44 Driving motor 45 ... Film formed on semiconductor wafer 46 ... Comparator, 47 ... Joint system 48 ... Driver, 49 ... Set light quantity 50 ... Photoelectric exchange 51 ... Slit diaphragm

Claims (2)

(57) [Claims] In a resist developing step, a development end point is determined by monitoring a change in intensity of reflected light with time of light irradiated from the upper surface of the resist, and a relationship between a predetermined finished dimension and a development end point; A developing method comprising adjusting a finished dimension by adjusting an exposure time based on a relationship between a development end point and an exposure time. 2. The developing method according to claim 1, wherein the relationship between the finished dimension and the development end point is represented by a polynomial, and the relationship between the development end point and the exposure time is represented by an exponential function.