JPH0580814B2 - - Google Patents

Description

Detailed Description of the Invention [Technical Field of the Invention] The present invention relates to a method for aligning two members, and in particular, the relative position between a mask and a wafer with reference to each mark provided on a photoelectric mask and a wafer. This relates to an alignment method for aligning.

[Technical background of the invention and its problems] Conventionally, when forming a resist pattern on the surface of a sample such as a silicon wafer, generally a fine width electron beam is scanned over the resist to expose the resist. This type of method has the problem that the scanning takes a long time, resulting in poor productivity. Recently, various types of transfer apparatuses have been developed that use a mask with a pattern formed in advance and transfer the mask pattern onto a sample using X-rays, electron beams, or the like. Among these transfer devices, a photoelectric mask that emits photoelectrons upon receiving ultraviolet light is used, and a magnetic field and an electric field are applied between this mask and the sample.
A photocathode mask type electron beam transfer device that performs transfer by focusing photoelectrons emitted from the mask is considered to be the most promising for fine pattern transfer.

FIG. 1 is a schematic diagram showing a photocathode mask type electron beam transfer device. The inside of the vacuum container (sample chamber) 1 is heated to 1×10 ^-6 by a vacuum pump 2.
It is evacuated to about [torr]. Vacuum container 1
A photoelectric mask 3 is placed at a predetermined position inside. This mask 3 is a quartz substrate 3 that transmits ultraviolet light.
a. A mask pattern 3b formed by attaching a thin film (for example, chromium) that blocks ultraviolet light to the lower surface of a quartz substrate 3a, and a C _S I or the like that emits photoelectrons upon receiving ultraviolet light is coated on the lower surface of these. It is formed from a photocathode 3c. Then, below the photoelectric mask 3, a sample 5 coated with a resist 4 is placed between the mask 3 and the sample 5.
They are arranged facing each other with a distance of about [mm]. Further, there is a light transmitting window on the upper wall of the container 1, and a transparent plate 6 is attached to close this window. Ultraviolet light from a light source 7 is introduced into the container 1 through the transparent plate 6, and the upper surface of the mask 3 is irradiated. On the other hand, on the outside of the container 1, for example, a Helmholtz coil 8 is provided.
The coil 8 applies a magnetic field in the vertical direction of the paper, that is, in the direction orthogonal to the photocathode 3c of the coil 8. Furthermore, a high voltage is applied between the mask 3 and the sample 5 from a DC power source 9, thereby applying an electric field in the same direction as the magnetic field.

The ultraviolet light emitted from the light source 7 is irradiated onto the upper surface of the photoelectric mask 3 through the transparent plate 6, whereby photoelectrons are emitted from the photocathode 3c of the mask 3 corresponding to the mask pattern 3b. The photoelectrons are focused and accelerated by the magnetic field and the electric field, travel downward, and enter the resist 4 on the sample 5. As a result, resist 4 becomes mask pattern 3.
The pattern is transferred by exposure according to the pattern b. In the figure, 10 is a shutter that blocks ultraviolet light from the light source 7, 11 is a drive system that drives the shutter 10, 12 is a coil power supply, 13 is a deflection coil, 1
Reference numeral 4 indicates a sample stage, 15 a drive mechanism, 16 a coil power supply, and 17 a drive power supply for driving the sample stage 14.

The relative positioning of the photoelectric mask 3 and the wafer 5 in this type of transfer apparatus is usually carried out by the following method, the basic principle of which is shown in FIG.
In advance, an alignment mark 21 for irradiating an electron beam by light irradiation is formed at a predetermined position on the mask 3, and an alignment mark 2 having the same shape as the mark 21 is formed at a predetermined position on the main surface of the wafer 5.
Form 2. The photoelectric mask 3 and the wafer 5 are placed facing each other, and the photoelectric mask 3 is irradiated with light and a magnetic field and an electric field are applied along the direction in which the photoelectric mask 3 and the wafer 5 face each other to generate an electron beam. The alignment mark 21 is irradiated onto the wafer 5. At this time, if the shapes of the marks 21 and 22 are lines of equal width arranged at equal intervals (pitch 2P) as shown in FIG.
(Since the generated X-rays are weak, such a pattern shape is used to increase the signal-to-noise ratio.) Relative positional deviation between mask and wafer and marks 21, 2
The relationship between the two and the overlapping area is as shown in FIG. That is, when the relative positions between the mask and the wafer are aligned, the overlapping area of the marks 21 and 22 becomes maximum, and the overlapping area decreases in proportion to the amount of positional deviation. Further, when the number of lines is large, similar changes in the overlapping area are repeated even if the amount of positional deviation exceeds ±P. By the way, when an accelerated electron beam bombards a metal target, the electrons emit part of their kinetic energy as X-rays as they decelerate due to collision with the target nucleus. The amount (intensity) of this X-ray generated varies depending on the electron energy, the atomic number Z of the target metal, and its thickness, but in general, Z
The larger the value, the greater the amount of X-rays generated. Therefore, the wafer 5 is made of Si, and the mark 22 is made of Ta, W.
When using heavy metals such as Mo, the electron beam
The amount of X-rays generated is different when irradiating the mark 22 and when the mark 22 is irradiated, and the amount of X-rays generated from the mark 22 is much larger than the amount of X-rays generated from the wafer 5. Therefore, the relationship between the X-ray output obtained from the X-ray detector 23 installed below the wafer 5 and the positional deviation amount is as shown in FIG.
By detecting the X-ray signal from this, the amount of positional deviation can be detected. For alignment, a deflection magnetic field is applied in the alignment direction by the deflection coil 13 and the coil power source 16, and the electron beam is
This is achieved by deflecting by the amount of positional deviation obtained by line signal detection. Instead of deflecting the electron beam, it is also possible to align the sample stage 14 by using the drive mechanism 15 and drive power source 17 to move the sample stage 14 by the amount of positional deviation.

By the way, in such a positioning method, since a detection signal with a good S/N ratio is required to obtain high accuracy, the following method is adopted. That is, by passing a current containing an alternating current component through the deflection coil 13, the electron beam is deflected in an alternating current (AC) manner. This electron beam modulation signal is used to perform synchronous detection (Lockin detection) of the output signal.
Performs wave detection, phase detection PSD) processing. At this time, a sine wave is selected as the AC waveform, and the amplitude (actually the beam scanning width) is set to the half pitch length of the line (=
P), the output after signal processing has output characteristics as shown in FIG. In other words, the fundamental frequency component PSD (1) is the second harmonic component PSD (a) in Figure 6a.
(2) is as shown in FIG. 6b. Then, by using these two signals, -P to +P
It is possible to detect the position up to For example, if the amount of positional deviation is x ₁ (-P < x ₁ < P), then the output of PSD (1) is y ₁ , and the amount of positional deviation is
Detected as x ₁ or x ₂ . On the other hand, PSD
Since the output of (2) is y ₂ , it is possible to distinguish between x ₁ and x ₂ . This can be similarly detected for positional deviations from -P to +P, so
By checking the outputs of both PSD(1) and PSD(2), position detection from -P to +P becomes possible. Therefore, alignment is performed by correcting the position so that the output of PSD (1) indicates zero at the center.

However, this type of method has the following problems. In other words, if the number of lines is large, even if the detection range from -P to +P is exceeded, PSD(1), PSD(2) within ±P
Since almost the same curve as the curve is repeated, for example, if the starting position is beyond ±P,
The problem is that the position is corrected to a position that is shifted by an integral multiple of the pitch, and it is impossible to determine whether it is at the correct position (center zero position). Therefore, if the position determined by pre-alignment between the sample and the mask exceeds ±P, the position will be corrected to an incorrect position. Recently, in order to achieve ultra-precise alignment, COARSE/FINE alignment, which consists of two stages of alignment, COARSE alignment using a coarse mark pattern and FINE alignment using a fine pattern, has been performed. A similar problem occurs if COARSE alignment does not ensure alignment within the pitch width of the FINE pattern.

There is no way to check whether there is a deviation of one pitch or more before transferring, and it can only be confirmed after transferring and developing at that position.
For this reason, unless the pre-alignment or COARSE alignment between the sample and the mask is performed to ensure that the alignment is within ±P, the image may be transferred to the wrong position on the sample, resulting in a decrease in productivity. Moreover, the number of samples processed in such a device is several tens per hour.
It cannot be ignored that in many cases pre-alignment or COARSE alignment suddenly exceeds ±P. Furthermore, if some trouble occurs in the pre-alignment mechanism due to long-term use and accuracy within ±P cannot be guaranteed,
For the above reasons, the number of defective products increases significantly, leading to a significant decrease in productivity. Such a decrease in productivity is a serious problem in mass-produced semiconductor manufacturing equipment.

[Object of the Invention] An object of the present invention is to be able to perform relative positioning of a first member such as a photoelectric mask and a second member such as a wafer with high precision using a synchronous detection method;
It is also an object of the present invention to provide a positioning method that can determine whether or not the position has been erroneously aligned beyond the position detection range, and that can contribute to improving the transfer throughput of an electron beam transfer device.

[Summary of the invention] The gist of the present invention is to improve the pattern shape of alignment marks formed on a photoelectric mask and a wafer (or sample stage), and to make it possible to determine whether or not there is incorrect alignment outside the detection range. It is to do so.

By synchronous detection as shown in Figure 6a and b above.
The reason why the PSD output is the same in the range of ±P and in the range exceeding this is because the lines of equal width as the marks are arranged at equal pitches.
Therefore, it is thought that repeating the same waveform can be avoided by arranging the lines at different pitches. That is, when the above-mentioned pitch is varied, the peak location of the synchronous detection output characteristic differs greatly between when the relative position of the photoelectric mask and the sample is within the detection range and when it is outside the detection range. Therefore, by comparing the second harmonic component when the position has been corrected with a preset reference value, it is possible to determine whether or not the detection range has been exceeded and alignment has been made to an erroneous position.

The present invention focuses on such points, and provides a first alignment mark on a first member such as a photoelectric mask for pattern transfer, and a second member such as a wafer used for electron beam transfer. A second alignment mark is provided on the member to emit a beam such as an X-ray by electron beam irradiation, the first mark is irradiated with light, and the beam emitted from the mark is irradiated onto the second member. At the same time, the beam is AC-deflected by an alternating current wave of constant amplitude, the beam emitted from the second mark is detected at this time, and the detected signal is synchronously detected to obtain its fundamental wave component and second harmonic component. , in an alignment method in which the amount of relative positional deviation between the above-mentioned members is determined based on a predetermined relationship from each of these components, and the relative position between the above-mentioned members is corrected according to this amount of deviation, each of the above-mentioned marks is set in plural numbers. This is a method in which the lines are formed and these lines are arranged at different intervals.

[Effects of the Invention] According to the present invention, when aligning a photoelectric mask and a wafer using a synchronous detection method, errors may occur in alignment prior to precision alignment such as pre-alignment, exceeding the detection range of precision alignment. Even if the image is aligned in the wrong position, it can be detected before the transfer is performed. Therefore, it is possible to prevent the image from being transferred at an incorrect position during pre-alignment, and the throughput of the electron beam transfer apparatus is greatly improved. Further, even if there is an abnormality in the pre-alignment mechanism or the transport mechanism, the abnormality can be quickly detected, so the operation of the transfer apparatus can be stopped before transferring to the wafer. This not only avoids a sudden drop in productivity, but also reduces the time required to start up the equipment.
Furthermore, significant effects can be obtained in terms of maintenance.

[Examples of the Invention] The details of the present invention will be explained below with reference to Examples. FIG. 7 is a plan view showing the pattern shape of the alignment mark used in the above embodiment.

In this embodiment, an electron beam transfer device having the configuration shown in FIG. 1 was used, and the following alignment marks were used. That is, the first alignment mark 21 formed on the photoelectric mask (first member) 3 is composed of a large number of lines as shown in FIG. 7, and the width of all the lines is 10.
It is equal to [μm]. And the spacing between adjacent lines is different, in this example the spacing is 10
[μm], 15 [μm], 20 [μm], 10 [μm], 15 [μm]
]、
20 [μm],... Further, the shape of the second alignment mark pattern on the wafer (second member) 5 corresponding to the above mark is also the same shape.

Using a mark with such a pattern shape,
FIG. 8 shows the relationship between the amount of positional deviation and the overlapping area when the mask and wafer are overlapped. This figure shows a cross section of the wafer-shaped mark 31 and the alignment electron beam 32.
In Fig. 8 a, b, c, and d, the positional deviation amount is 0 [μm],
These are the values at 10 [μm], 20 [μm], and 30 [μm].
FIG. 9 shows the characteristics of the overlapping area, and the output characteristics of the X-rays obtained from the detector at that time also have a shape similar to that shown in FIG. Note that the first mark line is formed of a thin film such as Cr, and the second mark line is formed of a heavy metal such as Ta or W.

Next, an actual positioning method will be explained. First, as in the case of transfer, an electron beam is generated using the photoelectric mask 3, and the beam is focused onto the wafer 5 by applying a magnetic field and an electric field. The alignment electron beam generated from the first mark 21 on the photoelectric mask 3 is focused onto the second alignment mark 22 on the wafer 5, and the amount of X-rays generated from the mark 22 at that time is 22, the X-ray output characteristics when the marks 21 and 22 are used are as shown in FIG. 9. Therefore, as described above, by passing a current containing an AC component through the deflection coil 13, an AC deflection magnetic field is applied and the electron beam is modulated. Using this electron beam modulation signal, the output signal is subjected to synchronous detection processing. That is, the modulated frequency component is extracted from the modulated output signal, and its fundamental wave component and second harmonic component are determined (this is set as the PSD output). At this time, select a sine wave as the modulation waveform, and set the amplitude (beam deflection width)
If we set 1/2 of the pitch, that is, 10 [μm], then
The output characteristics after synchronous detection are as shown in FIG. 10a for the fundamental wave component PSD(1) and as shown in FIG. 10b for the second harmonic component. Here, if we focus on the range within ±10 [μm], the shape is the same as the conventional position detection characteristic, and the conventional method can be applied as is. Therefore, the amount of positional deviation at the mask alignment position after pre-alignment or COARSE alignment is ±10
If it is within [μm], conventional alignment is performed. In the present invention, pre-alignment or
There is a problem with alignment when COARSE alignment exceeds ±10 [μm], and the following is an explanation of this case.

Now, suppose that pre-alignment is performed with a deviation of about 20 μm, for example, and the sample is sent to the mask alignment position. At this time, if the position is corrected so that the output of the fundamental frequency component (Figure 10a) of the output characteristic by the above-mentioned synchronous detection method approaches the zero point, it will be approximately 25 μm from the normal position (position of zero position deviation). ] Aligned to a misaligned position. The output of FIG. 10b, which is twice the frequency component at this time, exhibits a value β that is about 50% larger than the output α in the case of the normal position. Therefore, the value of γ, which serves as a judgment standard, is set in advance between α and β, and if the output of the double frequency component is smaller than γ, a signal indicating completion of alignment is given, and if it is larger than γ, pre-alignment is performed. Make sure to give a bad signal. Here, the amount of deviation in pre-alignment is now about 20 [μm], but if this amount exceeds ±10 [μm] and alignment is performed after the deviation is up to 65 [μm], the same applies to Double frequency components exhibit β values smaller than γ.
Therefore, it is possible to determine whether the prealignment is good or bad within the range of ±65 [μm].

Note that the sequence shown in FIG. 11 can be considered as a sequence in the case of prealignment failure. In other words, when a prealignment failure signal is given, the sample is sent to the prealignment mechanism without being transferred. Therefore, pre-alignment is performed again, and the mask is sent to the mask alignment position again. Then, similar position detection and alignment are performed again, and these operations are repeated until a signal indicating that alignment is complete is output. During this time, the number of times this rotation is performed is counted, and if the number exceeds a preset number, an abnormality signal is given and the positioning operation is terminated midway. By doing so, even if an abnormality occurs in the pre-alignment mechanism, transport mechanism, etc., it is possible to prevent the above-mentioned operation from being performed for a long time. The same applies to COARSE alignment, but in this case the pre-alignment
It only replaces COARSE alignment.

Thus, according to the method of this embodiment, a positional deviation of up to ±65 [μm] can be tolerated in pre-alignment or COARSE alignment, and the probability of alignment at an incorrect position is extremely small. Therefore, the throughput of the device is improved, and even if there is an abnormality in the pre-alignment mechanism or transport mechanism, the abnormality can be quickly detected, the number of samples transferred by mistake can be minimized, and the throughput can be improved. Not only can rapid deterioration be avoided, but the time required to start up the device can be shortened, and a great effect can be obtained in terms of maintenance.

Note that the present invention is not limited to the embodiments described above. In the above example, the line spacing is 10
[μm], 15 [μm], 20 [μm], 10 [μm], 15 [μm]
]、
Since it is 20 [μm]..., the tolerance range after pre-alignment or COARSE alignment is ±65
Although it is within [μm], it is possible to further widen the allowable range by changing this interval.
For example, the spacing is 10 [μm], 12 [μm], 14 [μm], 16
The spacing may be [μm], 18 [μm], 20 [μm], 10 [μm], etc., and the present invention is not limited to the above-mentioned spacing, and any spacing that can obtain the effect may be used. Further, the width of the line, etc. can be similarly deflected. Furthermore, the deflection amplitude of the beam can also be deflected as appropriate. Further, the member forming the second mark is
The material is not limited to Ta or W, but any material that emits X-rays by electron beam irradiation may be used. In short, the present invention can be implemented with various modifications without departing from the gist thereof.

[Brief explanation of the drawing]

Figures 1 to 6 are for explaining the conventional alignment method. Figure 1 is a schematic configuration diagram showing a photocathode mask type electron beam transfer device, and Figure 2 is a diagram of the basic principle for alignment. , Fig. 3 is a plan view showing a conventional alignment mark pattern, Fig. 4 is a characteristic diagram showing the change in mark overlap area with respect to the amount of positional deviation, and Fig. 5 is a diagram showing the change in the mark overlap area with respect to the amount of positional deviation. Characteristic diagram showing changes, Figure 6 is the 3rd
Characteristic diagrams showing the output characteristics of the signal after synchronous detection when the marks in the figure are used, FIGS. 7 to 11 are for explaining one embodiment of the present invention, and FIG. 8 is a schematic diagram showing the overlapping state of the mask and wafer, FIG. 9 is a characteristic diagram showing the change in X-ray output with respect to the amount of positional deviation, and FIG. A characteristic diagram showing the PSD output characteristics obtained by synchronously detecting the X-ray output in Figure 9.
FIG. 11 is a flowchart schematically showing the alignment sequence in the above embodiment. 1... Vacuum chamber, 2... Vacuum pump, 3... Photoelectric mask, 4... Resist, 5... Wafer, 6...
Transparent plate, 7... Light source, 8... Focusing coil, 9...
DC power supply, 10... Shutter, 11... Shutter drive system, 12... Focusing coil power supply, 13... Deflection coil, 14... Sample stage, 15... Drive mechanism, 1
6... Deflection coil power supply, 17... Drive power supply, 21
...first alignment mark, 22...second alignment mark, 23...X-ray detector.

Claims (1)

[Claims] 1. A first positioning mark is provided on a first member, and a second positioning mark is provided on a second member disposed opposite to the first member. , irradiating the first mark with light and irradiating the beam emitted from the mark onto the second member, and deflecting the beam with an alternating current wave of constant amplitude;
At this time, the beam emitted from the second mark is detected, the detected signal is synchronously detected, its fundamental wave component and second harmonic component are obtained, and the relationship between the above-mentioned members is determined based on a predetermined relationship from each of these components. In the positioning method of determining the amount of relative positional deviation and correcting the relative position between the respective members according to this amount of deviation, each mark is formed by a plurality of lines, and these lines are arranged at different intervals. , and each mark is formed in the same pattern. 2. The alignment method according to claim 1, wherein a sine wave is used as the constant amplitude alternating current wave.