JPH08128812A - Monitor for length measurement - Google Patents

Abstract

PURPOSE: To enhance the reliability of a monitor by a method wherein a line monitor and a hole monitor which correspond to an actual pattern are used and a pattern for recognition is used. CONSTITUTION: A monitor 13 for length measurement is formed on a scribing line 12. A gate polysilicon monitor GP, a first-layer metal monitor M and the like as line monitors are arranged in parallel at its lower stage so as to correspond to an actual pattern. In addition, a LOCOS monitor F, a first contact monitor C and the like as hole monitors are arranged in a matrix shape at the upper stage of the monitor 13 so as to correspond to the actual pattern. Consequently, it is possible to obtain the monitor 13 under a condition very close to the actual pattern, i.e., in which the proximity effect of a lithographic process and the microloading effect of an etching process are reflected. In addition, the respective monitors can be guided easily into the field of view of a length-measuring GEM by using L-shaped patterns 4, for recognition, at four corner parts of the monitor 13.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention has an object to improve a length measuring monitor for controlling a processing dimension of a semiconductor device.

[0002]

2. Description of the Related Art A conventional length measuring monitor will be described below with reference to FIG. In the manufacture of semiconductor devices, various patterns such as wirings and contacts are formed, but it is necessary to know how these patterns are actually formed. A natural and reliable method is to measure the pattern actually formed in the actual LSI. However, according to this method, the length must be measured for each individual LSI.
Since the number of length measurements is huge, the work becomes very complicated.

Therefore, on the scribe line (2) between the formation areas (1) of the LSI, a simulated pattern imitating a pattern actually formed on the LSI is formed and its dimension is measured. Has used a method for saving labor. According to this method, it is sufficient to provide several (5 to 6) simulated patterns on one wafer and measure the length thereof. Therefore, the work is much more difficult than the method of measuring the actual pattern. It can save labor. Such a simulated pattern is called a length measurement monitor. Usually, the shape of the length measuring monitor (3) is a single line shape as shown in FIG. 4, and in the figure, the LOCOS monitor (F), the gate / polysilicon monitor (GP), and the first contact monitor are shown. (C), first layer metal monitor (M), second contact monitor (SC), second layer metal monitor (S)
M), and so on.

Then, the line width of the pattern actually formed on the LSI is measured by measuring the resist pattern line width after the photolithography process and the pattern line width after etching of each monitor by a length measuring SEM (Scanning Electron Microsope). This is the same as obtaining the dimensions, and it is also possible to understand how much the difference between the dimensions of the resist pattern and the dimensions of the Al wirings, contacts, etc. formed on the basis of the dimensions.

[0005]

However, the above-mentioned conventional length-measuring monitor has the shape of a single line. For example, many wiring lines and contact holes are formed in an actual LSI such as a memory cell. However, since the above-described conventional length-measuring monitor is formed by a single line, light is not generated in the lithography process. In the proximity effect where the exposure state changes due to diffraction and interference due to adjacent patterns, and in the dry etching process,
As a so-called micro-loading effect, in which carbon atoms [C] scattered from the mask resist pattern affect the etching state of the adjacent pattern, the interaction caused by the proximity of the actual pattern can be monitored. Since it cannot be reflected, LS
There will be a considerable disparity with the actual pattern inside I, and it will not match the actual situation.

In particular, the difference increases relatively with miniaturization, and it becomes a disparity that cannot be ignored in submicron level products, and eventually the actual pattern has to be measured. On the other hand, if the length is measured using an actual pattern, a program for controlling the length measuring SEM must be created for each pattern of each chip on the wafer, and the program file becomes huge. Also,
Due to the increase in the number of program files, there have been problems such as an increase in the time for creating the program files and a complicated work.

[0007]

In order to solve the above problems, as shown in FIG. 1, a line monitor in which a plurality of lines corresponding to an actual pattern are arranged in parallel and a plurality of lines corresponding to the actual pattern are provided. By arranging the contact holes in a matrix form in a matrix for each corresponding manufacturing process on the scribe line, a monitor that reflects the actual pattern state of the semiconductor device can be obtained. Made it possible.

Further, in order to correspond to each of the above monitors,
For example, L-shaped recognition patterns are arranged at the four corners of the monitor.

[0009]

According to the length-measuring monitor of the present invention, as shown in FIG. 1, a plurality of lines corresponding to the actual pattern are arranged in parallel in place of the single-line monitor arranged on the scribe line. The line monitor and the hole monitor in which a plurality of contact holes corresponding to the actual pattern are arranged in a matrix are used. Therefore, as in a memory cell in which a large number of adjacent contact holes are arranged, To obtain a monitor that can reproduce conditions very close to the pattern in a pseudo manner and that reflects the proximity effect of the lithography process and the micro-loading effect of the etching process caused by the interaction of adjacent patterns in such an actual pattern. Will be possible.

Further, according to the length measuring monitor of the present invention, even if the miniaturization progresses, a monitor having a very small difference from the actual pattern of the LSI can be obtained, so that the reliability of the length measuring result is high. Better, eliminating the need to directly measure the actual pattern. Therefore, compared to the case where the actual pattern is directly measured, it is not necessary to create a program for controlling the length measurement SEM for each pattern of each chip on the wafer, and the program size can be significantly reduced. It is possible to prevent an increase in the time required for creating a program file due to an increase in the program size, and a complication of the length measurement work, which is accompanied with the increase.

Further, since the recognition pattern is arranged corresponding to each monitor, each monitor can be easily brought into the visual field of the length measuring SEM.

[0012]

EXAMPLE A monitor for length measurement explained in this example is shown in FIG.
As shown in FIG. 3, the wafer is formed on the scribe line 12 provided between the LSI formation regions (11) of the wafer,
As a line monitor (remaining pattern), a gate / polysilicon monitor (GP), a first layer metal monitor (M), a second layer metal monitor (SM), and a third layer metal monitor (TM) are arranged in blocks at the bottom. As a hall monitor (a blank pattern), a LOCOS monitor (F), a first contact monitor (C), and a second contact monitor (SC) are arranged in a block shape on the upper stage. As a whole, each of the above monitors is arranged. The length measuring monitors (13) are arranged in a matrix for each manufacturing process.

Further, L-shaped recognition patterns (14) are arranged at the four corners of the monitor so that each monitor can be easily guided within the visual field of the length measuring SEM. L
The character-shaped recognition pattern (14) is composed of two sets of patterns, the first recognition pattern 14A is formed of F, GP, C, M, and the second recognition pattern 14B is M, SC, S.
As in the case of M, TC, and TM, the remaining pattern and the blank pattern are alternately formed so as to reduce the step difference of the base. Thereby, the accuracy of pattern recognition can be improved.

As shown in FIG. 2, the line monitor comprises, for example, an Al layer in which five lines having a minimum line width are arranged in parallel. In addition, as shown in FIG. 3, the hole monitor is, for example, 25 contact holes with a minimum diameter arranged in 5 rows × 5 columns, and one minimum hole arranged at a position 5 μm to 150 μm away from them. It consists of a caliber contact hole.

According to the length-measuring monitor of the present invention, instead of the single-line monitor arranged on the scribe line, a line monitor in which a plurality of lines corresponding to the above-mentioned actual pattern are arranged in parallel is provided. In addition, since a hole monitor in which a plurality of contact holes corresponding to an actual pattern are arranged in a matrix is used, it is possible to use an actual hole monitor such as a memory cell in which many adjacent Al wirings and contact holes are arranged. The conditions very close to the pattern can be reproduced in a pseudo manner, and in such an actual pattern, the proximity effect of the lithography process caused by the interaction of the adjacent patterns and the micro
It is possible to obtain a monitor that reflects the loading effect.

Further, in the hole monitor, a single contact hole corresponding to the sparse state of the actual pattern is provided at a position distant from the above-mentioned plurality of contact holes. A pattern such as a peripheral circuit in a cell area or an ASIC can be reproduced in a pseudo manner, and a monitor corresponding to the pattern with a small proximity effect can be obtained. The measurement result of this monitor and the plurality of contacts can be obtained. By comparing with the measurement result of the length-measuring monitor in which the holes are arranged in a matrix, it is possible to know how much the diameter of the contact hole changes due to the proximity effect or the like. The single contact hole is provided at a position away from the plurality of contact holes in order to eliminate the proximity effect between the two.

Experimental results for explaining the working effects of the length measuring monitor of this embodiment are shown below. (1) Experimental Results of Line Monitor Table 1 shows measurement results of a conventional single line Al monitor and an Al line monitor (first metal layer M) in which five lines having the minimum line width of this example are formed. It is the table which compared and contrasted.

[0018]

[Table 1]

In Table 1, PE average is an average value of line widths of Al monitors, and PE3σ is a value obtained by multiplying a dispersion value σ of line widths of a plurality of Al monitors provided in a wafer by 3. is there. Further, the PR average means that a plurality of wafers are provided on the wafer.
The average value of resist line width of Al monitor, PE3σ
Is a value obtained by triple the dispersion value of the resist line width.
Further, the CD loss is the difference between the PR average and the PE average.

As shown in Table 1, the CD loss of the conventional monitor is -0.229 μm, which is a non-negligible difference at the submicron level, and the reliability of the monitor is significantly reduced. This is due to the proximity effect in the lithography process that occurs when the actual patterns are close to each other,
This is because the micro loading effect in the etching process is not reflected on the monitor.

However, in the monitor of this embodiment, the CD loss is -0.096 μm, which is considerably reduced as compared with the conventional one, and the error is negligible even at the submicron level. It was confirmed that the monitor was closer to the actual pattern than the monitor. Further, comparing PE3σ and PR3σ between the conventional monitor and the monitor of this embodiment, the monitor of this embodiment shows a lower value, and it can be confirmed that there is less variation depending on the location. Also, the monitor of this embodiment is closer to the actual pattern than the conventional one,
It can be said that it is a highly reliable monitor. (2) Experimental Results of Hall Monitor Table 2 is a table comparing and comparing the measurement results of the conventional single line hall monitor and the hall monitor (first contact C) of this embodiment. The hall monitor of this embodiment has 25 contact holes arranged in 5 rows × 5 columns corresponding to the dense state of the actual pattern and a single contact pattern corresponding to the sparse state of the actual pattern. The measurement results of the contact holes are shown.

[0022]

[Table 2]

In Table 2, the PE average means a plurality of hole monitors (usually the center of the wafer,
It is provided at five points: top, bottom, left, and right. ) Is the average of the aperture values after etching, and PER is the difference between the maximum and minimum aperture values of the plurality of hole monitors. Further, the PR average is an average value of aperture values after a photoresist process of a plurality of hole monitors provided in the wafer,
The PRR is the difference between the maximum and minimum values of the diameters of the plurality of contact hole monitors.

The CD loss is the difference between the PR average and the PE average, and is the difference between the size of the resist opening and the size of the hole monitor formed by etching using the resist as a mask. Shows. Regarding the CD loss, as shown in Table 1 in this experiment, there is a difference of 0.06 to 0.07 μm between the conventional monitor and the monitor of this embodiment, which can be ignored at the submicron level. This is a significant difference, and the reliability of the conventional monitor is significantly reduced.
Further, when comparing the PER and the PRR of the monitor of the conventional example and the monitor of the present example, the values of the monitor of the present example show smaller values, and from this point as well, the monitor of the present example
Is closer to the actual pattern than the conventional monitor,
It was confirmed to be a highly reliable monitor.

Further, the length measuring monitor of this embodiment, 25 contact holes arranged in 5 rows × 5 columns corresponding to the dense state of the actual pattern and the sparseness of the actual pattern. Comparing the measurement results of a single contact hole corresponding to each state, there is a difference of 0.009 μm in CD loss, a difference of 0.016 μm in PER and a difference of 0.002 μm in PRR. This is due to the proximity effect in the lithography process due to the density of the pattern and the micro roll in the etching process.
This is probably because it reflects the different Ding effect.

[0026]

As described above, according to the length measuring monitor of the present invention, a line monitor in which a plurality of lines corresponding to an actual pattern are arranged in parallel and a plurality of contacts corresponding to the actual pattern are provided. Since we are using a hole monitor that has holes arranged in a matrix, we can simulate conditions that are very close to the actual pattern, such as a memory cell with many adjacent Al wirings and contact holes. Since the monitor that can be reproduced and has a small difference from the actual pattern and the actual pattern state of the LSI can be obtained, the reliability of the measurement result is improved and the actual pattern is improved. Need not be measured directly.

Further, as compared with the case where the actual pattern is directly measured, it is not necessary to create a program for controlling the length-measuring SEM for each pattern of each chip on the wafer. As a result, it is possible to suppress the increase in the time for creating the program file, which has been caused by the increase in the program size, and the complication of the length measurement work, which is accompanied by the increase.

Further, since the recognition pattern is arranged corresponding to each monitor, each monitor can be easily brought into the visual field of the length measuring SEM.

[Brief description of drawings]

FIG. 1 is a top view illustrating a length measuring monitor according to an embodiment of the present invention.

FIG. 2 is a top view illustrating a line monitor according to an embodiment of the present invention.

FIG. 3 is a top view illustrating a hole monitor according to an embodiment of the present invention.

FIG. 4 is a top view illustrating a conventional length measuring monitor.

Claims (2)

[Claims] 1. A line monitor in which a plurality of lines corresponding to an actual pattern are arranged in parallel and a hole monitor in which a plurality of contact holes corresponding to an actual pattern are arranged in a matrix form on a scribe line. A length measurement monitor characterized by being arranged in a matrix for each corresponding manufacturing process. 2. The length-measuring monitor according to claim 1, wherein a recognition pattern is arranged corresponding to each monitor.