JPH01102304A - Measuring method of length by charged beam and apparatus therefor - Google Patents

Abstract

PURPOSE:To enable the highly precise measurement of a pattern width with excellent reproducibility and thereby to improve a yield in manufacture of LSI devices, by measuring the acceleration voltage dependence characteristic of secondary electron noise inherent in a pattern material and by conducting measurement with an acceleration voltage corresponding to the minimum value thereof. CONSTITUTION:A minimum value extraction circuit 22 extracts the minimum value of secondary electron noise and an acceleration voltage corresponding thereto from the acceleration voltage dependence characteristic of the secondary electron noise inherent to a pattern material. A psophometer 23 is a means to measure the amount of the secondary electron noise, and the input thereof is AC-coupled to the output of a secondary electron detector 10 through a capacitor 24. By this construction, an acceleration to be a specified measuring precision is selected from the acceleration voltage dependence property of the amount of noise of secondary electrons 9 obtained by scanning the surface of a sample 7 having a pattern on the surface by a charged beam 8, and from the characteristic of dependence of the measuring precision of a pattern width on the amount of the secondary electron noise. Then, from a secondary electron waveform obtained by scanning the surface of the sample by this acceleration voltage, the pattern width on the surface of the sample can be calculated in a pattern width calculation circuit 17.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a charged beam length measurement method and an apparatus therefor applied to pattern inspection during LSI manufacturing.

[Conventional technology]

BACKGROUND ART Charged beam devices represented by scanning scattering electron microscopes (hereinafter referred to as SM) have come to be widely used as non-contact and high-resolution surface observation means. A major field of application of such a charged beam device is considered to be a pattern width measuring device, which is a pattern inspection means during LSI manufacturing.

LSI mainly consists of an active element section that controls logic operations, etc., and an LS
In order to realize the function of I, it is composed of an electrode wiring part that organically connects these parts, and has a laminated structure in which each layer is formed by batch processing of eight sea urchins. That is, L.S.
I is manufactured by repeating a plurality of steps for forming each layer. Therefore, in order to ensure the target electrical characteristics and yield upon completion of LSI manufacturing, it is essential to perform a pattern inspection every time the pattern formation process (resist exposure, development, etching, etc.) is completed.

In such LSI pattern inspection, since it is required to be able to observe the surface with high precision without damaging the sample, SEM has come to be used exclusively in place of the conventional optical microscope.

Figure 7 shows the general configuration of a pattern width measurement (hereinafter referred to as length measurement) device using an SEM. In this figure, 1 is an electron gun, 2 is a condenser lens, 3 is an objective lens, 4 is an aperture, and 5
1 is a deflector, 6 is a stage, 7 is a measurement sample such as an LSI device (hereinafter referred to as a sample), 8 is an electron beam emitted from the electron gun 1, and 9 is a sample when the sample 7 is irradiated with the electron beam. 10 is a secondary 'C' detector, 11 is a DC voltage source for the electron gun 1, 12.
13 is a condenser lens 2. a direct current source that excites the objective lens 3; 14 a scanning circuit that controls the deflector 5;
5 is a stage control circuit that controls the movement of the stage 6;
Reference numeral 6 denotes a console, and by manually operating the console, the accelerating voltage of the electron beam 8 is set via a DC voltage source 11, and the accelerating voltage of the electron beam 8 is set via a DC voltage source 11, and each condenser lens 2. The excitation of the objective lens 3 is adjusted to set the beam current value of the electron beam 8 and to focus the electron beam 8. Set the irradiation position and scanning conditions (direction 2 width, speed). 1T is a pattern width calculation circuit, 18 is a display such as a CRT provided in the pattern width calculation circuit, and 19 is a control device for the pattern width calculation circuit (Japan Society for the Promotion of Science Industrial Application of Charged Particle Beams No. 13).
2 Committee, 85th Study Group Materials, pl, 1983).

Next, the procedure for pattern length measurement will be explained using FIG. First, by operating the console 16, the electron beam 8 repeatedly scans the same pattern on the surface of the sample 7, that is, the surface of the LSI device, and the secondary electrons 9 emitted thereby are used to calculate the pattern width through the secondary electron detector 10. A circuit 1T stores a secondary electron waveform that changes in accordance with the shape of the pattern and is averaged. The line width of the pattern can be determined by displaying an SEM image of the pattern using the secondary electron waveform on the CRT 18, placing a cursor on the edge of the pattern, and calculating the distance between the cursors by the pattern width calculation circuit 17, or by using a threshold value. is specified, and the line segment connecting the intersection with the rising part and the falling part of the secondary electron waveform is defined as the line width of the pattern, and this is calculated by the pattern width calculation circuit 17 (hereinafter referred to as line profile). law).

[Problem that the invention seeks to solve]

In the length measuring device mentioned above that measures the pattern dimensions of LSI,
Since it is necessary to collect large amounts of minute data, high speed and high accuracy are required.

To this end, it is desirable to optimize not only the system configuration but also the measurement conditions from the above viewpoint. Measurement conditions include (a) charged beam irradiation conditions, (a) secondary electron waveform smoothing conditions, and (e) pattern width calculation conditions, but from the viewpoint of measurement accuracy, -) and (b) are is important. ~), in addition to the above-mentioned averaging, moving average, fast Fourier transform, etc. are known, and their effects on measurement accuracy have been experimentally confirmed. On the other hand, regarding (&), the current situation is that the accelerating voltage conditions, which are considered to be highly dependent on the pattern material to be measured, are selected empirically depending on the pattern material, and the measurement accuracy Almost no quantitative relationship has been clarified. For this reason, measurement accuracy varies depending on the skill level of the measurer and the pattern material to be measured.
There was a problem with the reliability of pattern dimension management during manufacturing.

Therefore, the present invention has been made in view of the above-mentioned conventional problems, and its purpose is to improve the yield in LSI device manufacturing by making it possible to measure pattern widths with good reproducibility and high precision. An object of the present invention is to provide a beam length measurement method and a device thereof.

[Means for solving problems]

In order to solve the above-mentioned problems, it is sufficient to have a method and means for quantitatively extracting the accelerating voltage of the charged beam that satisfies the target measurement accuracy according to the material of the pattern to be measured.

In the charged beam length measurement method according to the present invention, the amount of secondary electron noise superimposed on the secondary electron waveform obtained by scanning the sample surface with a charged beam is the dominant factor in measurement accuracy, and the total yield on the sample surface is (The sum of the secondary electron emission ratio and the backscattering coefficient) This was done by focusing on the fact that the secondary electron noise is proportional to the sum of the secondary electron emission ratio and the backscattering coefficient. The main feature is that it achieves high accuracy in length measurement by measuring with voltage.

The charged beam length measuring device according to the present invention includes means for accelerating, converging and deflecting the charged beam, means for changing the acceleration region pressure while keeping the beam current of the charged beam constant, and collecting secondary electrons from the sample surface. means for determining the shape of the sample surface using a secondary electron waveform, means for measuring the amount of secondary electron noise, and means for calculating the minimum value of a function of one variable that changes within a certain range. ing. Alternatively, means for measuring the absorbed current may be provided in addition to the means for measuring the amount of secondary electronic noise.

[Effect]

In the present invention, it is possible to measure the pattern width of an LSI device that is in the process of being manufactured with great precision.

〔Example〕

Embodiments of the present invention will be described in detail below with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of a charged beam device for explaining an embodiment of the charged beam length measurement method according to the present invention, and the same parts as those in the previous figures are denoted by the same reference numerals. be.

In the figure, 21 is a computer, and this computer 21 and a DC voltage source 11 constitute a means for changing the accelerating voltage of the electron beam 8, and a DC current source 12 and a DC current source 1
3 constitute means for keeping the beam current of the electron beam 8 constant. 22 is a minimum value extraction circuit for extracting the minimum value of secondary electron noise and the corresponding acceleration voltage from the accelerating voltage dependence characteristic of secondary electron noise specific to the pattern material; 23 is a means for measuring the amount of secondary electron noise; At a certain i-front needle, its input is AC coupled to the output of the secondary electron detector 10 via a capacitor 24. The computer 21 includes a DC voltage source 11, a DC current source 12, a DC current source 13. Scanning circuit 14. stage'
Control circuit 15. It is connected to the pattern width calculation circuit 1T and the minimum value extraction circuit 22, and in order to centrally control these, it organically executes processing such as transmission and reception of control signals and data, and data processing and storage.

FIG. 2 is a block diagram of a charged beam device for explaining another embodiment of the charged beam length measurement method according to the present invention. In the figure, numerals 1 to 22 indicate the same parts as in FIG. 1. 25 is a current needle that irradiates the sample 7 with an electron beam and measures the absorbed current flowing from the ground to the sample at 9 o'clock.

FIG. 3 shows the dependence of the measurement accuracy on the amount of secondary electronic noise. In the figure, 30 is the photoresist AZ pattern on Si (pattern width = 1 μm), and characteristic 31 is the pattern of pair S1 ( This is an actual measurement example of pattern width = 1 μm).

Fig. 4 shows the accelerating voltage dependence characteristics of the root mean square value of the secondary electron noise amount and the total yield when the pattern material is Mo, and Fig. 5 shows the accelerating voltage dependence of the total yield when the pattern material is 810z. Characteristics 32 and 34 are actual measurement examples of the accelerating voltage dependence characteristics of the root mean square value of the amount of secondary electron noise, and characteristics 33 and 35 are actual measurement examples of the accelerating voltage dependence characteristics of the total yield.

Next, an embodiment of the charged beam length measurement method based on the accelerating voltage dependence characteristic of the amount of secondary electron noise will be explained using FIG. 1 and FIGS. 3 to 5. In FIG. 1, a computer 21 is connected to a DC voltage source 11. DC current source 12° DC current source 13. Scanning circuit 1
4. Stage control circuit 15. The electron beam 8 irradiation condition signal, pattern width calculation conditions, etc. are transmitted to the pattern width calculation circuit 1T and the minimum value extraction circuit 22, and the acceleration voltage Vp is changed to scan a specific area on the surface of the sample 7 with the electron beam 8. . In parallel with this, the secondary electrons 9 emitted from the specific area are collected by the secondary electron detector 10, and the root mean square value n of the AC noise component of the secondary electrons (hereinafter n is the secondary electron noise f[ ) is measured by the noise meter 23 via the capacitor 24.

Since the alternating current noise of secondary electrons is white noise, its root mean square value n is given by the following formula [Noise Analysis, Miyawaki-O, p. 35, Breakfast Shoten (1967)].

n2=2+1・bow g@1B (1) Here, e is the electron charge, Ig is the signal current collected by the secondary electron detector 10, and jB is the secondary electron detector 1G, the capacitor 24, and the noise meter 23. This is the bandwidth of the detection system composed of. Since the signal current Ig is proportional to the amount of secondary electrons from a specific region of sample 7 in Fig. 1, n2 = Ksu(Vp)'' Ip (2) Here, K is a constant that does not depend on the surface material of sample 7. , ρ(Vp
) is sample T when irradiated with electron beam 8 at accelerating voltage Vp.
The total yield is That is, if the acceleration voltage vp is varied while the beam current Ip is kept constant and the root mean square value n of the secondary electron noise amount is measured, the total yield ρ(Vp) specific to the pattern material can be calculated.
An accelerating voltage dependence characteristic proportional to is obtained.

Characteristic 32 in FIG. 4 and characteristic 34 in FIG. 5 are actual measurement examples measured in this manner. When the acceleration voltage Vp is in the range of 0.8 kV to 2.5 kV, the pattern material is M.

In the case of FIG. 4, the amount of secondary electronic noise decreases monotonically with the accelerating voltage vp, and becomes the minimum at Vp = 2.5 kV.

On the other hand, in the case of 8102, the amount of secondary electronic noise becomes minimum at VP=1 kV, as shown in FIG.

Now, the M3 diagram shows the dependence of the measurement accuracy σ on the amount of secondary electron noise n when the pattern width is measured using the line profile method mentioned above, and the measurement accuracy is the standard deviation value σ of the pattern width measurement value obtained by repeatedly measuring the measurement accuracy. Show characteristics. Actual measurement examples of the photoresist AZ pattern on S1 of characteristic 30 and the bare S1 pattern of characteristic 31 demonstrate that the measurement accuracy σ increases approximately in proportion to the 315th power of the amount of secondary electron noise n. That is, when the pattern material is No, the measurement accuracy σ becomes minimum when Vp=2.5 kV, and when the pattern material is No. 8102, the measurement accuracy σ becomes minimum when Vp=1 kV. The minimum value extraction circuit 22 in FIG. 1 selects an accelerating voltage that provides good measurement accuracy from the accelerating voltage dependence characteristics of the amount of secondary electron noise that differ depending on the pattern material, as shown in the characteristic 32 in FIG. 4 and the characteristic 34 in FIG. 5. The extraction circuit can be constructed using a normal arithmetic circuit.

Next, another embodiment of the charged beam length measurement method based on the accelerating voltage dependence characteristic of the total yield will be described using FIGS. 2, 4, and the rJs diagram. This case was made by focusing on the fact that the root mean square value n2 of the amount of secondary electron noise is proportional to the total yield ρ(Vp) specific to the pattern material, as can be seen from the equation ψ). That is, from the accelerating voltage dependence characteristic of the total yield ρ(Vp),
If the accelerating voltage corresponding to the minimum value of the total yield a(Vp) is extracted, for the reason mentioned above, the total yield #(VP) that can minimize the measurement accuracy σ can be calculated using the absorption current of the sample T as 15(Vp). Then, it is given by the following formula.

p(Vp) = 1+Il(vp)/1Ipl
(3) Here, the sign of the absorption current l5(Vp) is assumed to be positive when the absorption current l5(Vp) flows from the sample T in FIG. 2 to the ground. Therefore, the absorption current l5 (Vp) is measured with the ammeter 25 in FIG. 2, and the corresponding total yield p
(Vp) and its minimum value may be calculated by the minimum value extraction circuit 22. Characteristic 33 in Figure 4 and qIf property 35 in Figure 5
This is an example of the acceleration voltage dependence of the total yield measured in this manner. From both figures, the acceleration voltage dependence characteristics of the total yield of both Mo and 5102 are in good agreement with the acceleration voltage dependence characteristics of the root mean square value of the amount of secondary electron noise, and therefore, the acceleration that minimizes the amount of secondary electron noise is It can be confirmed that the voltage is the same as the accelerating voltage that minimizes the overall yield.

FIG. 6 shows an actual measurement example of the accelerating voltage dependence characteristic of the measurement accuracy when measuring the length of a mono wiring pattern with a line width of 1 μm at an accelerating voltage in the range of Vp=1 to 2 (kV). In the same figure, the characteristic 36 is the number of times of averaging of the secondary electron waveform M=
4. Characteristic 3T is a dependent characteristic under the condition of M=16, and characteristic 37 is a dependent characteristic under the condition of M=64.

From the figure, it can be seen that in any case, the measurement accuracy depends on the accelerating voltage, corresponding to the accelerating voltage dependence characteristic of the root mean square value of the amount of secondary electron noise in characteristic 32 in Fig. 4, or the accelerating voltage dependence characteristic of the total yield in characteristic 33. It can be confirmed that there is a monotonous decrease with The amount of change in the volume of the secondary electron beam when the accelerating voltage is changed can be extracted from Figure 4 when the pattern material is Mo, and the amount of change in measurement accuracy corresponding to this amount of change can be extracted from 315 in Figure 3. It can be predicted by the power law. The amount of change in measurement accuracy when changing the accelerating voltage shown in FIG. 6 corresponds relatively well with the predicted value obtained by such a procedure. That is, FIG. 6 proves that the present invention is effective.

In the above-mentioned embodiments, a length measuring device using an electron beam is mainly explained, and it goes without saying that the same method and method can be used for other charged beams such as an ion beam.

〔Effect of the invention〕

As explained above, according to the present invention, highly accurate pattern width measurement can be performed with good reproducibility, so an extremely excellent effect can be obtained that can contribute to improving the yield in manufacturing L81 devices.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing one embodiment of the charged beam length measuring device according to the present invention, FIG. 2 is a block diagram showing another embodiment of the charged beam length measuring device according to the present invention, and FIG. 3 is a block diagram showing one embodiment of the charged beam length measuring device according to the present invention. Figure 4 is a diagram showing the dependence of measurement accuracy on the amount of secondary electron noise. Figure 4 is a diagram showing the dependency of the accelerating voltage on the root mean square value of the amount of secondary electron noise and the total yield when the pattern material is Mo. The pattern material is 5t.
Figure 6 shows the root mean square value of the secondary electron noise amount and the accelerating voltage dependence characteristics of the total yield in the case of o2, and the pattern material is M. FIG. 7 is a diagram showing the acceleration voltage dependence characteristic of measurement accuracy in the case of 5E3A. 1... [child M, 2-... condenser lens, 31
1...Objective lens, 4...Aperture, 5II.
...Deflector, 6...Stage, T...Sample, 8
...Electron beam, 9...Secondary electron, 1G-...
Toyo secondary electron detector, 11 am DC voltage source, 12.1
3-・φ・DC current source, 14・・・Scanning circuit, 15・
...Stage control circuit, 1B--Console,
17...Pattern width calculation circuit, 18...Display device,
19...control device, 21...fist calculator, 22...
・・Minimum value extraction circuit, 23・・・Noise meter, 24・・Skewer/capacitor, 25・・・Ammeter, 30.31・・0
・Accelerating voltage dependence characteristic of measurement accuracy, 32.34... Accelerating voltage dependence characteristic of root mean square value of secondary electron noise amount, 33,
35...-acceleration voltage dependence characteristics of total yield, 36-3B-
...Acceleration voltage dependence characteristics of measurement accuracy. Patent applicant Nippon Telegraph and Telephone Corporation

Claims (3)

[Claims] (1) Accelerating voltage dependence of the amount of secondary electron noise or total yield obtained by scanning a charged beam on the surface of a sample with a pattern, and the amount of secondary electron noise or total yield of the pattern width measurement accuracy Select an accelerating voltage that provides a specified measurement accuracy based on the dependence characteristics on the sample, and measure the pattern width on the sample surface using a secondary electron waveform obtained by scanning the sample surface with the accelerating voltage. Characterized charged beam length measurement method. (2) means for accelerating, converging, and deflecting a charged beam on a sample surface having a pattern on the surface; means for changing an accelerating voltage while keeping the beam current of the charged beam constant; and a secondary beam emitted from the sample surface; means for collecting electrons; means for calculating the pattern width on the sample surface from the waveform of the secondary electrons; means for measuring the amount of secondary electron noise; A charged beam device comprising: means for calculating. (3) means for accelerating, converging, and deflecting a charged beam on a sample surface having a pattern on the surface; means for changing the accelerating voltage while keeping the beam current of the charged beam constant; and secondary energy emitted from the sample surface; means for collecting electrons; means for calculating the pattern width on the surface of the sample from the waveform of the secondary electrons; means for measuring the absorbed current flowing into the sample; A charged beam device comprising: means for calculating.