GB2257504A - Determining the relative position of semiconductor water patterns - Google Patents

Abstract

To determine the position of a mask pattern formed on a surface of a semiconductor wafer (1) relative to the position of a reference pattern previously formed on the surface, a photosensitive film (8) is applied to the surface, and areas of the film are selectively irradiated with light via a photo mask to form a latent image of the mask pattern of exposed and unexposed areas of film. The reference and mask patterns include diffracting portion (4) and (5) respectively. Relative movement is effected between the wafer and a laser beam (6) so that the diffracting portions (4) and (5) of the reference and mask patterns are successively irradiated by the beam. Laser light is diffracted by the reference portion (4) and also by the mask portion (5) due to the difference between the indices of refraction of the exposed and unexposed parts of the latent image prior to development thereof. The spacing between the resulting diffraction patterns is measured to determine the spacing (x) between the reference and mask patterns. <IMAGE>

Description

METHOD OF MEASURING RELATIVE POSITIONING ACCURACY OF A PATTERN TO BE FORMED ON A SEMICONDUCTOR WAFER BACKGROUND OF THE INVENTION Field of the Invention The present invention relates to a method of determining the position of a mask pattern formed on a surface of a semiconductor wafer relative to the position of a reference pattern previously formed on that surface.

Description of the Related Art Semiconductor integrated circuits are now being made with enhanced operating speeds and refined pattern geometries. The refinement of the patterning geometry owes much to the advancement of the technology employed in their fabrication.

In the microfabrication process, the technique for improving the accuracy of alignment of successive patterns, along with the technique for producing fine patterns, is crucially important.

Generally, the maximum misalignment allowed in the prior art was about 20% of the minimum of all of the dimensions involved. Accordingly, the alignment needs to be improved along with the fine patterning as the latter is improved.

The methods for aligning wafer steppers which are currently most important in exposure devices include a method which utilizes a television image, a method which utilizes a laser beam, and the like.

In the former case, as disclosed in the U.S. Patent No. 4,640,619 granted to K W Edmark III, for example, alignment marks are formed in a layer of photoresist by irradition with ultraviolet rays and the marks are observed with an optical microscope or video camera by ultra violet rays. In this method it is of course possible to detect the alignment marks after developing the photoresist. On the other hand, it is also possible to detect the alignment marks by means of its latent images formed by exposure without ever developing the photoresist.However, it should be noted that this later method has been used only for setting an initial condition of the wafer stepper or for correcting the characteristic of a lens by detecting the difference of the contrasts in an exposed part and an unexposed part, and has so far never been proposed in a method for the detection of the mask-pattern positioning accuracy.

An example of the latter method has been published, in Proc.

SPIE, Vol. 538 (1985), pp. 9-16. As shown in Fig. 1, the principle of the method is to form an alignment mark 14 on a semicondcutor substrate 1. The mark 14 is composed of seven square segments, each of which is a concavity or convexity on the surface of the substrate.

The seven segments are aligned. The substrate 1 is mounted on a stage (not shown) whose position is perfectly controlled, and the stage is moved through a He-Ne laser beam 6 which is maintained in a fixed position. As shown in Fig. 1, the laser beam 6 has an extended ellipsoidal cross-section. As the stage moves through the beam 6, light of the laser beam is reflected from the substrate 1 and the segments in the mark 14 act as a diffraction grating. Maximum intensities of reflected light are given at angles On given by the n equation P sin En = n;, where P is the period of the grating n segments and is the wavelength of the light. The position of the stage is therefore determined from a detection of the position at which a maximum intensity of reflected light is obtained.

For greater accuracy a plurality of grating lines may be provided. The diffraction signals from these lines are then detected successively, stored in a memory and, upon completion of scanning, the individual positions of the marks are determined. The average of the individual positions is then calculated and treated as the aligned position of the plural mark.

Now, the factors that determine the accuracy of alignment of a wafer supported on a wafer stage of a wafer stepper may roughly be divided into two kinds, namely, those due to the performance of the stepper itself, such as the stage accuracy and the distortion of the projecting lens, and those due to the accuracy of alignment of the lithographic processes used in fabrication of the semiconductor devices formed on a wafer.

Of these factors, the measurement of the accuracy of performance of the stepper itself is a measurement of the error between a designed spacing and the actual spacing between maskpatterns or the spacing between the latent images formed by the exposures alone. In determining the alignment accuracy of the latter type, a measurement is taken of the error between a designed value of the spacing and the actual spacing between the pattern of a lower layer of material and a developed photoresist pattern which is overlaid on the lower layer pattern.

In particular, for the measurement of the positioning error between a lower layer pattern formed in a pre-process and a photoresist pattern formed subsequently, it has been conventional to use an optical microscope or a measuring device exclusively designed for this purpose.

The conventional method for evaluating the performance of a wafer stepper by detecting a latent image in a photoresist possesses a fundamental problem in that discriminating between an unexposed region and an exposed region becomes increasingly difficult with increased irradiation times, due to the fact that ultraviolet rays are used for detection. Namely, in a method in which ultraviolet rays are applied to the photoresist latent image, detection is made possible by detecting the difference in the indices of refraction and reflectance of an exposed part and an unexposed part for an ultraviolet light source. However, the indices of refraction and the reflectance of the unexposed part are changed when ultraviolet irradiation is employed for detection, resulting in a small signal difference or poor reproducibility.Moreover, due to irradiation with ultraviolet rays, there is another problem in that when the semiconductor wafer needs be developed later on, the region which was previously exposed for the purpose of alignment will have to be developed.

On the other hand, a method of detecting diffraction gratings by laser beams involves a complicated procedure of positioning the wafer by detecting the diffraction grating images formed in a pre-process, exposing the semiconductor wafer to be inspected, performing a development process in a developing unit after removing the wafer temporarily from the wafer stepper, and then inspecting the wafer using an appropriate inspection device. The time it takes before obtaining the inspection result and the man hours required therefor result in a significant loss in production efficiency.

SUMMARY OF THE INVENTION It is a main object of the present invention to provide a method of measuring the accuracy of the position of a mask pattern formed in a photoresist wherein the measurement is effected quickly after exposure and enhances the production efficiency.

According to the invention there is provided a method of determining the position of a mask pattern formed on a surface of a semiconductor wafer relative to the position of a reference pattern previously formed on that surface, characterized by applying a photosensitive film to the said surface; selectively exposing at least one area of the film to irradiation by a light such as ultra violet rays by means of a photo-mask, thereby to form a mask pattern of exposed and unexposed film; effecting relative movement between the wafer and a He-Ne laser beam so that the reference pattern and the mask pattern are successively irradiated by the laser beam; and detecting light of the laser beam diffracted by the reference pattern and light of the laser beam diffracted by the difference between the indices of refraction of the exposed and unexposed parts in the mask pattern, prior to development thereof, thereby to determine the spacing between the reference pattern and the mask pattern.

A preferred method in accordance with the present invention utilizes the principle that when a photoresist on a semiconductor wafer is exposed to light such as ultra violet rays with a sensitizing wavelength, the index of refraction of the exposed parts of the photoresist is changed with respect to that of unexposed parts, due to a photochemical reaction induced by the irradiation. If the exposed pattern forms a diffraction grating, it is possible to obtain diffraction images by irradiating the exposed regions in the direction perpendicular to the semiconductor wafer with a He-Ne laser.Then, it becomes possible to measure the relative position of the optical latent image of the diffraction grating transferred to the photoresist, which has a reference diffraction grating formed therein or which is formed on a semiconductor wafer having a reference diffraction grating formed thereon, by detecting the diffracted light of the laser to measure the spacing between the diffraction gratings.

The present invention is aimed at performing all of the series of processes described in the above by using a reduction stepper.

BRIEF DESCRIPTION OF THE DRAWINGS The above and further objects, features and advantages of the present invention will become more apparent from the following detailed description of embodiments taken in conjunction with the accompanying drawings, wherein: Fig. 1 is a plan view showing the position of a He-Ne laser beam and a diffraction grating in a prior art method; Fig. 2(a) is a plan view of a part of a semiconductor wafer formed with diffraction gratings in a method according to the invention; Fig. 2(b) is an enlargement of a part of Fig. 2(a) which further shows a He-Ne laser beam used in detecting the gratings; Fig. 2(c) is a section on the line A-B of Fig. 2(b); Fig. 2(d) is a waveform diagram showing the signal intensity of light diffracted in the method of Fig. 2(a);; Fig. 3 is a section of a semiconductor wafer showing the generation of diffracted light by a diffraction grating in the form of an optical latent image formed in photoresist on the wafer, the section illustrating the principle of the present invention; Fig. 4(a) to Fig. 4(c) show an embodiment of the present invention, Fig. 4(a) being a sectional view showing one stage in a fabrication process and Fig. 4(b) and Fig. 4(c) being the waveforms of the intensity of the diffracted light; and Fig. 5(a) to Fig. 5(c) show another embodiment of the present invention, Fig. 5(a) and Fig. 5(b) being sectional views illustrating the fabrication processes and Fig. 5(c) being a waveform showing the variation in the intensity of the diffracted light with location.

DESCRIPTION OF THE PREFERRED EMBODIMENTS Referring now to Fig. 2 of the drawings, Fig. 2(a) is a plan view of a semiconductor wafer 1 mounted on a movable wafer stage (not shown). The wafer 1 includes a large number of chips 3, each containing integrated circuits. The spaces between the chips form scribe lines 2. Formation of the integrated circuits on the chips 3 involves a series of fabrication processes, some of which are carried out whilst the wafer 1 is on the movable wafer stage of a stepper (a step-and-repeat apparatus on a reduction projection exposure apparatus) where the method of the present invention is applied and some of which are carried out after the wafer has been removed to external apparatus.

As mentioned above, accurate positioning of the wafer 1 involves two alignment processes. First, the wafer 1 must be accurately positioned by moving the wafer stage within the wafer stepper ie. the wafer must be accurately positioned with respect to components such as the optical projection system, hereinafter described. Secondly, once the wafer is accurately positioned, a pattern formed on the wafer 1 by exposure to radiation via a mask in the wafer stepper must be accurately aligned with respect to any patterns formed in previous processes.

At the beginning of the fabrication process a first layer of photoresist (not shown) is formed on an upper surface of the wafer in a suitable photoresist coating equipment. The wafer 1 is then mounted on a wafer stage of a wafer stepper.

The wafer stage is moved to a predetermined position in the wafer stepper and a series of patterns 4 is formed in the first layer of photoresist. Each pattern 4 consists of a series of spaced, square regions, each region being approximately 4 x 4 microns and the spacing (p) between the centres of adjacent regions being approximately 8 microns. The patterns 4 of the first photoresist, which are not visible at this stage, are formed of unexposed regions of photoresist, the remaining surface area of the first layer of photoresist being exposed to ultraviolet light via a suitable reticle or mask.

The wafer 1 is now removed from the wafer stepper, the first layer of photoresist is developed to remove all of the previously exposed areas thereof, and the wafer is then subjected to etching.

Semiconductor material is thus removed from exposed areas of the surface of the wafer 1, leaving each of the patterns 4 as a series of square convexities 7 on the upper surface of the wafer. A second layer of photoresist is now applied to the surface of the wafer, forming a coating over the originally exposed areas and over the patterns 4 of the convexities 7 as shown in Fig. 2(c).

The patterns 4 form a grating having a pitch p, as shown in Fig. 2(b) of the drawings.

The wafer 1 is now returned to the wafer stage of the wafer stepper. To ensure that the wafer 1 is accurately located in its previous position relative to the optical projection system and other components in the stepper, the surface of the wafer is irradiated by a laser beam. The pattern 4 of the convexities 7 then acts as a diffraction grating and the position of the wafer 1 is located by detecting the maximum intensities of radiation reflected from the pattern. The detected position of the wafer is compared automatically with the predetermined inital position and the position of the wafer stage is automatically adjusted to bring the wafer to that predetermined position.

To ensure that geometrical patterns subsequently formed on the surface of the wafer 1 are correctly positioned, a second diffraction grating is now formed in the second photoresist layer 8. This second grating is formed by irradiating the surface of the wafer 1 with ultra violet rays via a second mask to form a latent pattern 5. The second mask is formed with series of square apertures, one series for each chip on the wafer. After removing the second mask, each region of the second photoresist layer 8 including the latent pattern 5 is exposed to the laser light as shown in Fig. 2(b). The aligned regions in each pattern 5 of the latent image in the second layer of photoresist extend parallel with and are spaced a distance XA from the regions in an associated pattern 4 of the convexities 7 of the semiconductor material.

It will be appreciated that the regions in each pattern 5 are regions of photoresist which have been exposed to ultra violet rays but which have not been developed. Each pattern 5 forms a diffraction grating defined by latent images in the second layer of photoresist.

Fig. 3, which is a section taken lengthwise of a pattern 5, is used to describe the principle by which light of laser beam is diffracted from an optical latent image of the diffraction grating formed on a semiconductor wafer 1 and consisting of exposed parts 2 and unexposed parts 3 of photoresist.

If n1 represents the index of refraction of a part 2 of the photoresist exposed with ultra violet rays, and n2 represents the index of refraction of an unexposed part 3 of the photoresist, a He-Ne laser beam (wavelength# / = 633mm) propagates with a wavelength A1 = A/n1 within the part 2 and a wavelength A2 = A/n2 within the part 3. When the laser beam 10 is incident upon the photoresist from above, the beam is reflected at the interface between the photoresist and the semiconductor wafer 1 and comes out back into the atmosphere.The primary reflected light I1 coming out of each exposed part 2 and the reflected light I2 coming out of each unexposed part 3 have a phase difference, and the pattern formed by the parts 2 and 3 forms a diffraction grating with a pitch p.

In this case, the intensity of the diffracted light has a maximum S/N in comparison to the scattered light when the thickness t of the photoresist satisfies the relation

that is, when the phase difference between I1 and I2 is equal to an odd multiple of A/2.

On the other hand, when the phase difference equals an integral multiple of A. namely, when there is satisfied

the intensity of the diffracted light has a minimum S/N in comparison to the scattered light so that the diffracted light is hardly distinguished from the scattered light.

Further, the angle e between a line perpendicular to the surface of the wafer 1 and a line along which the diffracted lights strengthen each other is determined by the equation P.sin8 = k\, (k = O, 1, 2, 3, ...), .... (3) The detector 10 has to be installed therefore at a position corresponding to the direction of the angle 8.

Referring again to Fig. 2 of the drawings, Fig. 2(c) is a section taken along the line A-B of Fig. 2(b). Fig. 2(c) shows convexities 7 in the first pattern 4 and an exposed part 5 in the second pattern 5 in the second layer 8 of the photoresist.

Due to misalignment, each pattern 5 may be displaced from the desired position relative to the associated pattern 4. Accordingly, to determine the relative positions of the patterns 4 and 5, the wafer stage is moved across the axis of an optical projection system which generates a He-Ne laser beam 6. As shown in Fig. 2(b), the beam 6 has a cross-section in the form of an elongate ellipse whose major axis extends parallel with the gratings forming the patterns 4 and 5.

During movement of the wafer 1 through the beam 6, the intensity of light reflected from the surface of the wafer varies in the manner shown in Fig. 2(d), there being a first peak in intensity when the pattern 4 moves through the beam 6 and a second peak when the pattern 5 moves through the beam. The spacing XA between the two peaks is equal to the spacing XA between the patterns 4 and 5 and is measured by a control system of the stepper.

The above process of determining the relative positions of associated patterns 4 and 5 is repeated for a selection of patterns distributed over the surface of the wafer 1.

Data relating to the spacing of the associated patterns is analysed in the control system of the stepper. From this analysis, information regarding a misalignment of the patterns 5, relative to the patterns 4 is obtained. The control system then acts upon this information to move the wafer stage into an optimum position wherein there is a minimal displacement of each pattern 5 and associated chip 3 from the desired position relative to the optical projection system.

With the wafer 1 in this optimum position, each of the chips 3 is now exposed to a pattern of radiation via further mask or reticle.

This pattern defines areas of the chip which are to be subjected to etching, ion implantation, etc., in further processing.

The wafer 1 is removed to an external apparatus where this further processing is carried out. If it is then necessary to return the wafer to the wafer stepper, the wafer must be located so that any geometrical patterns of exposure, etching, etc., are effected in required positions of alignment with patterns already formed on the wafer. To this end, further patterns corresponding to the patterns 5 are formed in a layer of photoresist on the wafer and the wafer is then adjusted to bring this further pattern into a predetermined position relative to the pattern 4, in the manner described above.

Fig. 4 illustrates in more detail a further method according to the invention for determining the accuracy of alignment between a lower layer pattern formed in a former semiconductor fabrication process and a pattern in the form of an optical latent image in a photoresist. This method is one in which a diffraction grating pattern (corresponding to the diffraction grating 4 in Fig. 1) which serves as a reference for the alignment is formed by convexities 40 of polysilicon having a thickness of about 4000 A, there being an oxide film 49 with a thickness of about 300 A below the polysilicon. Fig.

4(a) shows the manner in which a second diffraction grating pattern 42 is formed by irradicating ultra violet rays through a mask or reticle 48 at a position spaced by a predetermined distance xA from a polysilicon convexity 40, the second diffraction grating pattern 42 being formed by exposure in a layer 43 of photoresist applied to a semiconductor wafer 41.

After the mask 48 is removed, when the semiconductor wafer 41 of the form shown in Fig. 4(a) is irradiated with a He-Ne laser beam and scanned, the intensity of the signal 71 derived from light diffracted by the diffraction grating pattern 42, consisting of an optical latent image in the photoresist 43, is much weaker than the intensity of the signal 72 derived from light diffracted by the polysilicon convexities 40, as illustrated in Fig. 4(b). It is important of course that the film thickness of the photoresist layer 43 satisfies Eq. (1) in the flat region where the latent image of the diffraction grating pattern 42 is formed.Since, however, the mere fulfilment of the above condition cannot solve the aforementioned problems, it becomes necessary to have a signal processing system which is capable of independently adjusting the intensities of the signals 71 and 72 generated by the latent image 42 and the polysilicon convexities 40, respectively.

Moreover, in order to more accurately determine the distance XA between two diffraction gratings made of different materials, it is necessary to execute the scanning by the He-Ne laser beam continuously and at once. Accordingly, in the system for detecting the diffracted light it is extremely important to automatically control independently of each other the signals obtained from the two diffracted lights of different intensities. For example, in Fig.

4(b), which shows the manner in which the intensities of the diffracted light signals vary with the scanning position of the He-Ne laser beam, a signal level adjustment is executed independently for positions on the negative side and the positive side, respectively, of an intermediate "0" position between the two diffraction gratings, and signal processing is effected so as to bring the respective signal levels to lie within a desired range of intensities.

As a result, after the signal processing, the electrical signals 71, 72 derived from the diffracted lights are made to assume equivalent levels despite the difference between the two diffracted light intensities, as shown in Fig. 4(c). This makes it easy to determine the relative positions of the diffracted light signals. A difference xM between the distance xA separating the two signal positions and the designed value xD of that distance, xM = xA XD, gives the accuracy of alignment between the polysilicon convexities 40 and the exposed pattern 42 of the photoresist 43.

Figs. 5(a) and 5(b) show sectional views of a wafer at different stages during a further method of measuring the alignment accuracy in accordance with the present invention and Fig. 5(c) shows the waveform diagram of the diffracted He-Ne laser light. In this method, the accuracy of the position of a second diffraction grating pattern 22 relative to a first diffraction grating pattern 21 is measured. First, the first diffraction grating pattern 21 (corresponding to the diffraction grating 4 in Fig. 1) is formed in a layer of photoresist 13 by a first time exposure by ultra violet rays using a first reticle 18. Then, the second diffraction grating pattern 22 (corresponding to the diffraction grating 5 in Fig. 1) is formed at a position spaced by a designed distance XA from the first pattern 21 by using ultra violet rays through a second reticle 18'.

In this case, it is necessary to design the reticles 18, 18' so as to expose the patterns 21, 22 and to leave unexposed those regions other than the two diffraction gratings 21 and 22.

Although there is not formed in this state a three-dimensional pattern in the photoresist 13 on the semiconductor wafer 1, an optical latent image is formed by the photochemical reaction in the photoresist 13 in which only the exposed diffraction grating patterns 21 and 22 have an index of refraction nl and the unexposed portions of the photoresist 13 have an index of refraction n2. Then, after removing the mask 18' by irradiating the photoresist vertically from above with a He-Ne laser beam 6\ = 633 nm) and scanning the wafer, diffracted light is generated at the positions of the diffraction patterns 21 and 22. Therefore, by plotting the intensities of the diffracted light against the position of the semiconductor wafer 1 there is obtained signal waveforms 71, 71' whose intensity is shown in Fig. 5(c).

When one derives the distance xA between the two peaks of the signal waveforms 71, 71' of the diffracted light and then computes the deviation xM of this distance from the designed value xD, namely, XM = XA xD, the result xM represents the alignment accuracy between the two times of exposure shots.

As explained in the above, it is possible to determine the position of the wafer by detecting diffraction by the optical latent image, by irradiating the optical latent image with a He-Ne laser beam. This eliminates the necessity for the developing of the pattern. Moreover, it is possible, in a wafer stepper having an alignment system which utilizes the diffracted light of a He-Ne laser beam, to measure the accuracy of the alignment on the spot, without removing the semiconductor wafer from the exposure stage after exposure of the photoresist. Accordingly, the availability of the exposure device can be enhanced substantially, which in turn will enormously affect the production efficiency of the semiconductor integrated circuit devices.Moreover, the frequency of masking required in the fabrication of the recent VLSIs is large, so that it becomes possible to save the time which is required for a single process of masking multiplied by the above-mentioned masking frequency, which also contributes to a reduction in the time taken for processing. Furthermore, a series of processes involving removal from the exposure stage, development and inspection of the semiconductor wafer can be eliminated so that there is an extremely significant consequential effect, such as a reduction in wages and salaries, not to mention an improvement in production efficiency and a reduction in the processing time.

Moreover, the conventional use of ultraviolet rays for detecting the alignment which can be avoided and there is absolutely no possibility of a change in the index of refraction of the unexposed portion of the photoresist with the length of time of irradiation, so that it is possible to enhance the accuracy, reproducibility and reliability of the position measurement. In addition, when the detection is conducted, the semiconductor substrate is irradiated with the He-Ne laser beam and not by ultraviolet rays. Accordingly, even if it is given a development processing after the measurement, the pattern on the semiconductor wafer will not be affected at all.

Furthermore, even if the difference between the intensities of the signal from the alignment mark formed in the initial process and the signal from the alignment mark due to the exposed but undeveloped latent image is significant, the difference between the signal intensities can automatically be controlled so that the respective signals can be positively detected by the series of stage scanning.

It will be appreciated that the above-mentioned patterns in the form of diffraction gratings can be replaced by other patterns which result in the diffraction of light of the He-Ne laser.

Claims (4)

CLAIMS:

1. A method of determining the position of a mask pattern formed on a surface of a semiconductor wafer relative to the position of a reference pattern previously formed on that surface, characterized by applying a photosensitive film to the said surface; selectively exposing at least one area of the film to irradiation by a light by means of a photo-mask, thereby to form a mask pattern of exposed and unexposed film; effecting relative movement between the wafer and a He-Ne laser beam so that the reference pattern and the mask pattern are successively irradiated by the laser beam; and detecting laser light diffracted by the reference pattern and last light diffracted by the difference between the indices of refraction of the exposed and unexposed parts in the mask pattern, prior to development thereof, thereby to determine the spacing between the reference pattern and the mask pattern.

2. A method as claimed in claim I, wherein the relationship between the thickness t of said photosensitive film, the index of refraction nl of said exposed part of said photosensitive film, the index of refraction n2 of said unexposed part of said photosensitive film, and the wavelength A of said He-Ne laser beam satisfy the relation

(k: positive integer)

3. A method as claimed in claim 1, wherein said mask pattern and said reference pattern are diffraction grating patterns each having a plurality of parts which are aligned with a constant pitch.

4. A method as claimed in claim 3, wherein said He-Ne laser beam extends perpendicularly of said photosensitive film and a detector for said diffracted light is installed at a position which satisfies the relation p sin 8 = where p is said constant pitch of said diffraction grating patterns, A is the wavelength of said He-Ne laser beam, 8 is the angle between the direction of said He-Ne laser beam and the direction of the diffracted light, and k is a positive integer.