JPH0574683A - Method for aligning reduction stepper - Google Patents

Abstract

PURPOSE:To accurately align a reduction stepper without lowering its throughput by performing die-by-die exposure only on wafers on which nonlinear deviations occur in such a way that the die-by-die exposure is performed when remaining errors become larger than allowable values and global exposure is performed in accordance with measured results when the remaining errors are lower than the allowable values. CONSTITUTION:An aligning exposing device measures the position of the shot of a wafer as fine alignment after performing pre-alignment and search alignment. Each correcting amount of shift, scaling, rotation, and orthgonality which causes a deviation from ideal coordinates is calculated from the measured results of the position. At the time of calculation, remaining errors Rx and Yy which are not included in the correcting amounts are also calculated simultaneously. When the errors Rx and Yy are lower than allowable values, global exposure is performed with the correcting amounts and, when the errors exceed the allowable value, die-by-die exposure is performed. Therefore, alignment can be performed with high accuracy without lowering the throughput.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a reduction projection exposure apparatus which is a semiconductor manufacturing apparatus, and more particularly to an alignment (superposition) method.

[0002]

2. Description of the Related Art FIG. 2 shows a conventional alignment method for a reduction projection exposure apparatus. In addition, the linear misalignment component is shown in FIGS.
Shown in. The wafer is initially O
Align F, search alignment in X direction,
Wafers are aligned in the Y direction and the rotation direction. Next, the position of an arbitrary shot is measured by fine alignment, and a linear shift component shift (Sx, S
y), scaling (Mx, My), orthogonality (θx-θy), and rotation (θy) are calculated. In die-by-die exposure (die-by-die alignment), exposure is performed while measuring the position for each shot thereafter. In global exposure (enhanced global alignment), exposure is performed by predicting the position of each shot based on (shift, scaling, orthogonality, rotation) calculated by fine alignment.
These two methods are prepared and used by being distinguished in advance according to the required alignment accuracy.

In FIGS. 3 to 7, the dotted line is the ideal position and the solid line is the position where the deviation occurs.

[0004]

As the miniaturization progresses, it has become necessary to improve the alignment (superposition) accuracy. In the conventional enhanced global alignment, position measurement is performed for each shot and exposure is performed. Therefore, in the case of a non-linear shift caused by warp or distortion of the wafer, there is a drawback that the alignment cannot be performed accurately.

Further, in the die-by-die exposure, since the position is measured for each shot, there is a drawback that the throughput is lower than that in the enhanced global alignment.

[0006]

According to the alignment method of the present invention, after measuring the position of an arbitrary shot, it is determined whether the deviation is a linear deviation or a non-linear deviation, and it is automatically switched to the residual. If the error exceeds the allowable value, die-by-die exposure is performed.

[0007]

According to the above construction, since the die-by-die alignment exposure is performed only on the wafer in which the non-linear shift has occurred, the throughput is improved and the alignment can be performed with higher accuracy than the die-by-die alignment exposure for all the wafers.

[0008]

Embodiments of the present invention will be described below.

FIGS. 7 and 8 show the direction of wafer deviation, and FIG. 7 shows a wafer in which a linear deviation occurs.
Indicates a wafer in which a non-linear shift has occurred (in the figure, the arrow indicates a vector in the shift direction, and the circle indicates a fine alignment position measurement shot). As shown in FIG. 1, the mesh exposure apparatus measures the position of the shot marked with ◯ on the wafer as fine alignment after pre-alignment and search alignment. Based on this measurement result, each correction amount of shift, scaling, rotation, and orthogonality that causes a deviation in ideal coordinates is calculated. At this time, at the same time, the residual error R not included in the above
x and Ry are also calculated. If Rx and Ry are below a certain allowable value, global exposure is carried out with a correction amount, and if exceeding the allowable value, die-by-die alignment exposure is carried out. As a result, the wafer of FIG.
The wafer is subjected to die-by-die alignment exposure, which has the advantage of not lowering the alignment accuracy.

[0010]

As described above, according to the present invention, by setting the allowable value of the residual error as described above, the die-by-die alignment exposure is performed only on the wafer in which the non-linear deviation occurs, so that the alignment can be performed accurately without lowering the throughput. .

[Brief description of drawings]

FIG. 1 is a flowchart of an alignment method of the present invention.

FIG. 2 is a flowchart of a conventional alignment method.

FIG. 3 is a diagram illustrating a shift component of a shift.

FIG. 4 is a drawing for explaining a displacement component of scaling.

FIG. 5 is a drawing for explaining a rotation deviation component.

FIG. 6 is a drawing for explaining a deviation component of orthodoxness.

FIG. 7 shows a wafer with a linear shift.

FIG. 8 shows a wafer having a non-linear shift.

Claims (2)

[Claims] 1. In a reduction projection exposure alignment (superposition) method, at the stage of position measurement (fine alignment) of an arbitrary shot in a wafer, it is determined whether a linear misalignment or a non-linear misalignment is detected, and automatically. A reduced projection exposure alignment method characterized by performing exposure while performing position measurement for each shot (die-by-die exposure) for wafers that have a non-linear shift, and performing global exposure for other wafers. 2. A residual error is calculated when determining a linear or non-linear deviation, and if the residual error is within a permissible value, global exposure is performed by correction, and if it exceeds the permissible value, the diver is exposed. Reduction projection exposure alignment method for performing red eye alignment exposure.