JPS58117405A - Pattern measuring method - Google Patents

Abstract

PURPOSE:To compute the size of the width of a pattern based on the angle formed with each cursor, by approximating the values of a plurality of positions obtained from an edge part, among the values of the positions corresponding to the width measuring positions to the values forming a straight line by a least square method. CONSTITUTION:The cursors K1 and K2 are moved to approximately intermediate points E and F of the lower part of the pattern. These positions are specified as width size position of the mark. The following operation is automatically performed. The specified positions of a point Po and cursors K1 and K2 are automatically read by the electronic computer 4. Relations between the specified positions and the peak to peak value a/b and a'/b are computed and memorized. Then the scanning position P1 is scanned at a minute interval. Curves theta and theta' formed by coordinate values from the edge part are approximated to a straight line by the least square method. The angle theta and theta' formed by the resulting straight lines and the cursors K1 and K2 are computed, and the product of the value of the size and costhetac is computed. The vaue is made to be the width size of the pattern at the scanning position.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a pattern measurement method using electron beam scanning.
In particular, it measures the width of a pattern that has an angle with respect to a reference direction and whose edges are not straight.

In recent years, integration has progressed from ICs to LSIs and ultra-L8Is, and patterns have become smaller accordingly. Now, when drawing a pattern on a wafer or mask blank, as the pattern (actually the width of the pattern) becomes smaller, the process technology for creating the pattern on the wafer, for example, the gas concentration during etching (plasma etching) becomes smaller. ) will also change. Therefore, in order to determine the creation process conditions according to the pattern dimensions (pattern width dimensions),
In addition, in order to develop a manufacturing process technology that corresponds to miniaturized pattern dimensions, it is necessary to know the shape of the pattern, essentially the width dimension of the pattern.

By the way, generally drawn patterns have a hem S, as shown in Figure 4, so when measuring the width dimension of the pattern, measure between the top and appropriate points of the hem from L to hem. Since the value differs depending on which distance between the boundary points of the flat area is measured,
It is necessary to decide in advance which one to measure.

Therefore, a new pattern width measurement method has been devised that allows measurement of the width from anywhere to anywhere in the pattern at the same time. In this method, an image of the pattern is scanned by an electron beam, and the image of the pattern is converted into, for example, Y
Two cursors in the direction and one marker in the X direction are displayed on the cathode ray tube screen, and as shown in FIG. Move the cursor Kl to the position Rf, F on the pattern to be measured as the width dimension (approximately the midpoint of the hem S in the figure).

Among the signals detected by line scanning on the wrinkle-specified position, the relation value between the signal corresponding to the carpet specified position and the peak value is calculated and stored, and thereafter, by scanning on the pattern, the relationship value is calculated and stored. This is a method of measuring the distance between specified positions in each scan from the detected signal and the relationship value.

Now, in the above method, the pattern displayed on the cathode ray tube screen has an angle with respect to the reference direction (for example, the Y direction when measuring the width of the pattern in the X direction), that is, with respect to the cursor. In this case, the cursor is aligned with the pattern edge number by rotating the wafer stage or by rotating the scanning direction of the beam. However, all of these methods incorporate individual differences among the measurers, resulting in large errors even in minute measurements. Furthermore, as mentioned above, the pattern generally has a hem, and since the hem is not linear but uneven, individual differences among the measurers are more likely to occur.

The present invention has been made in view of these points, and the image of a mosquito pattern is displayed on a display device by scanning the pattern with an electron beam, and the width dimension is measured at an appropriate position on the bottom of the pattern. A position to be measured is specified as , and a relationship value between a signal corresponding to the specified position and a peak value is calculated and stored among the signals detected by line scanning at the specified position, and then, From each signal detected by scanning a plurality of scanning positions in the vicinity of the scanning position where the pattern width dimension is to be measured and the relationship value, the distance between the specified positions missed in each scanning is calculated, and the width dimension measurement is performed. Among the position values corresponding to the positions, a straight line is formed using the method of least squares using the plurality of position values obtained from one edge part of the pattern and the plurality of position values obtained from the other edge part. The present invention provides a novel pattern measuring method in which the angles formed by each of the straight lines and each of the cursors are determined by approximating the value, and the width dimension of the pattern is calculated based on these angles. .

FIG. 3 shows an example of an apparatus implementing the pattern measuring method of the present invention.

In the figure, 1 is an electron gun, and the electron beam emitted from the electron gun is focused onto a focusing lens 2i (a wafer 3 on which a pattern has been drawn in a previous step).
The wafer 3 is scanned by a deflector 6 that operates in accordance with a scanning signal from a scanning signal generating circuit 5 that operates in response to a command. Reflected electrons generated from the wafer 3 due to the scanning number are detected by a reflected electron detector 1. The output of the detector is sent to the electronic computer 4 via the amplifier 8, and at the same time, the cathode ray is supplied with an output (scanning signal) synchronized with the output from the scanning signal generating circuit 5 to the deflector 6. It is also sent to tube 9. The cathode ray tube displays, for example, two cursors Kl and Km in the Y direction on the screen by its brightness modulation.
You can now put out a single marker M in the vX direction, and go to the right.
Each cursor and marker can be moved in the X and Y directions by external operations. Further, the position of the marker M on the anode ray tube is stored by the electronic computer 4, and the distance between the two cursors is calculated by the electronic computer. The reason why two cursors are displayed in the Y direction and one marker in the X direction as described above is to measure the width dimension of the pattern in the X direction. , displays two cursors in the X direction and one marker in the Y direction. Naturally, the number of cursors and markers is not limited to the above number, regardless of which direction the width dimension is to be measured.

Now, the reference scanning direction (for example, the X direction) as shown in FIG.
Next, a method of measuring the pattern width dimension at the scanning position 21 of the pattern finish forming an angle with respect to the pattern and having unevenness at the bottom edge will be explained.

First, the marker M is brought to the scanning position P of the pattern ending displayed on the screen of the cathode ray tube 9 by an external operation. Then, move the cursor Ks to the position where the mark width dimension is to be measured.
, Ks. In FIG. 4, the cursors Kg and Kl are respectively moved to the approximate midpoint JP of the bottom portion d of the pattern, and these positions are designated as mark width dimension positions. All subsequent operations are performed automatically (electrically). The scanning position P and the specified position of the cursor Kl, Kl are determined by the electronic computer 4.
is loaded into. The computer calculates the scanning position P of the electron beam.
, the position C on the signal of the signal sent from the detector 7 via the amplifier 8 (see FIG. 5(a)) relative to the specified position on the picture plane of the cathode ray tube, The corresponding level value 1.1' and the level value at a representative location of the signal sent from the detector T, for example, the peak-to-peak value, are detected, and the specified position and the corresponding peak peak value are detected. The relationships a/b and a'/b are calculated and stored. Next, the inside of the scanning position Ps is shifted by minute intervals and scanned. That is, the scanning position pH, Plm within the 21 parts
, @e@, Fun, and PAll are scanned according to instructions from the computer 4. First, when scanning pu, if a signal as shown in FIG. 5(b) is input from the detector 7 to the computer 4, the computer detects the peak-to-peak value of the signal, In addition to this, the relationship value a/
The products am and a'm of b and a'/b are calculated.

These products am and a- correspond to the specified mark width measurement position at the scanning position Pu of the pattern P. Then, the calculator calculates the value 11. Position 0 on the signal where jl was obtained
．． Measure the distance lx between b and use this measurement value as the scanning position P-
Assume that the width dimension of the pattern P is This distance 11
The scanning start point is the origin, and the detection positions Cm and I on the signal are
The X coordinates of k are set as CI and da, and a computer calculates the difference between the coordinates. In this way,
Computer number ζ scanning position Pus, @..., Rm,
By each scan of Pan, each scan position pu, ga @
@ , Psm, Psn The product corresponding to the specified mark width measurement position by 1 cm am, m, * *, t
Coordinates c of the position where n, an(, am, an' are obtained
m.

Measure and memorize d@, am fist, cm, dm, co, dn and board width dimensions 1@, @@・, 1m, In. Then, perform the following calculation.

First, as shown in FIG. 6, each scanning position Pu, Pa.

..., Psm, coordinate values Ct from one edge part of the pattern for Pall, cm, ..., c
m, cn and the coordinate values da, d- from the other edge part
. . , dm, and dn form curves θ and θ′, respectively, which are linearly approximated by the method of least squares. T and A calculate the straight line obtained as a result. This average angle -? This is the angle of the 1% scanning position of the pattern finish with respect to the reference direction (X direction). Also, each scanning position Pu, Pms, * em ,
ram, board width dimension in Pan, h,...
1m, the average value of In is taken as the width dimension of the scanning position R section in the reference direction. Then, the product of the dimension value j· and aSa is calculated, and this value is set as the width dimension of the pattern at the scanning position 21. Furthermore, when measuring the width dimension of another position JI Pa part of one pattern, the inside of the Ps part is sequentially scanned by shifting minute intervals in the same way as above, and the above relationship a/b e a'
/b to measure the product (substantially a counter value) and plate width dimension corresponding to the specified mark width dimension measurement position at each scanning position Psm, Pa..., and store these;
The width dimension of the pattern at position P is calculated by the calculation as described above.

In the above embodiment, the position on the signal where the value a weight is obtained is C.
In addition to I and Dt, there are also C'S and DI, but in advance, enter the inner position (
CI, DI) to be memorized to measure the distance between them.

Further, in the above embodiment, the position at which the mark width dimension is measured is specified by the cursors Kl, Kl, but by specifying one position with any one of the cursors, the level value i'' on the signal corresponding to this specified position is determined. The relationship a7b between and the peak-to-peak value b may be used as the relationship value of the mark width dimension designation positions (two positions).

According to the present invention, when measuring the width dimension of a pattern, it becomes clear from where to where the width dimension is being measured, and the width dimension at a unified designated position is automatically measured at each location of each pattern. It can be measured at high speed and with high precision, and especially when measuring the width of a pattern that has an angle with respect to the reference direction and whose edges are not in a straight line, it does not take into account the individual differences of the measurer, so accurate measurement results can be obtained. can get.

[Brief explanation of the drawing]

Fig. 1 is a cross-sectional view of a pattern, Fig. 2 is a diagram to supplement the explanation of the new pattern width dimension measurement method, Fig. 3 shows an example of a device implementing the pattern measurement method of the present invention, and Fig. 4
Figure 6 is a diagram to supplement the detailed explanation of the present invention. 3: Wafer, 4: Electronic computer, 6: Deflector, 1: Backscattered electron detector, S: Cathode ray tube, Kg, Zhen: Force-Sol, M
: marker, 1%m, Rm, e #@, plrll
l, pHl: scanning position p, N @pattern, P・
: Scanning position %L: Peak peak value, 11: Distance. Patent applicant JEOL Ltd. representative Tadao Kase, upper r r; Semuro-21-

Claims (1)

[Scope of Claims] The image of the pattern is displayed on a display device by scanning the pattern with an electron beam, and the position to be measured as the width dimension is specified on an appropriate position of the bottom of the pattern. Among the signals detected by line scanning on the specified position,
The relationship value between the signal and the peak value corresponding to the specified position is calculated and stored, and then each of the detected values is The distance between designated positions in each scan is calculated from the signal and the relationship value, and among the position values corresponding to the width dimension measurement position, the values at multiple positions obtained from one edge portion of the pattern and the other are calculated. Approximate each of the values of a plurality of positions obtained from the edge part to the value forming a straight line by the method of least squares, find the angle between each of the straight lines and the measurement direction, and calculate the angle between each of the straight lines. A method of measuring a pattern in which the width dimension of the pattern is calculated based on the angle.