DE2143737A1 - PHOTO ETCHING - Google Patents

Description

BobIIngen, August 13th Li) 71 gg-ba / sz

Ai.itLiches reference number: New registration Applicant's file number: Docket GE 9 71 006

Photo-etching process

The invention relates to a Photoät ^ process in which the surface a suitable material coated with photo lacquer, which exposes the photoresist through a photomask containing the pattern and developed and then carried out the etching of the iJeiterial will.

Such photo-etching processes are known to be used in Photolithography. An additional, newer field of application is semiconductor technology.

The method according to the invention is to be described particularly with regard to the application in semiconductor technology, without that thereby a restriction to this special field of application is intended.

All manufacturing processes currently customary in integrated circuit technology are based on the use of masked etching technology aligned. It is made of light-sensitive, organic Made use of compounds known as photoresists. In principle, by means of a

309810/0917

tolack used photomasks on a semiconductor wafer clLe various for diffusion metallization and other processes required pattern applied. The photomask used contains the photographically on the surface of a suitable one Mask material to be reproduced image patterns. For example, the upper curses could be used as materials in the individual processes of semiconductor material itself, and of covering the semiconductor material Oxide and metal layers in question.

It essentially contains what is known from photolithography | Follow the steps below.

The relevant one to be used as a hash in subsequent processes The surface of a suitable material is first coated with the photosensitive photoresist. On this photoresist a photomask containing the desired image pattern is applied and in particular on already in previous processes used image patterns aligned. After more suitable, through After exposure of the photoresist layer has taken place through the photomask, the image of the photomask is developed in the photoresist layer. Depending on the use of positive or negative photolich are either the exposed or the unexposed Parts of the photoresist layer removed. The remaining TaL-F The photoresist layers are fixed so that they can be removed from the following Etching and similar processes are not attacked. In particular, they form an etching mask with the help of which Image patterns predetermined by the photomask used in the Surface of the material arranged underneath can be transferred. The material is only etched in the Areas instead of where the photoresist layer is missing. After the etching process has been carried out, the remaining parts of the Photoresist layer removed so that the desired pattern is available for the subsequent processes.

The image pattern to be formed in the photoresist layer cannot be more perfect than the photographic pattern used for the reproduction.

^:) .c - !: et GE 971 OGf

309810/0917

2U3737

mask. For this reason, the quality of the photomask is essential. Of course they have to
geometric dimensions of the image pattern must be correct and these dimensions must be relatively independent of temperature and humidity influences. The image pattern must have a uniform density in both the light and dark areas
The transitions between light and dark areas must be sharply defined.

It has been shown in the past that conventional photo-etching techniques, especially when used in integrated circuit technology, are not as perfect as would actually be necessary with regard to an economically justifiable yield of integrated circuits that can be produced therewith. In order to achieve sharp lines and edges in the image pattern of the mask to be produced, efforts are made to select the thickness of the photomask used and the thickness of the photoresist layer as thin as possible. However, it cannot be ruled out that imperfections in the material and in the coating occur more frequently in thin layers and are just as exposed and reproduced as the actual mask image. In this case, defects in the photoresist layer usually consist in the fact that these areas are without photoresist
having. In the case of defects in the photomask, a distinction must first be made as to whether the photoresist is positive or negative,
So around photoresist, in which exposed or during development
unexposed photoresist is removed. Basically, however
to divide the defects in the photomask into bright defects
in inherently dark areas, hereinafter referred to as light defects,
and in dark flaws in areas that are actually bright, hereinafter referred to as dark flaws. All of the flaws considered here have in common that they only occur in a random distribution.

A photo-etching process is already known from US Pat. No. 3,317,320, which prevents the effects of imperfections in the photoresist layer. First of all, a first photo

Docket GB 971 006 309810/0917.

_ 4 -

lacquer layer exposed via a first mask, the photoresist layer developed and the corresponding parts removed. A second photoresist layer is then applied, which is exposed through a second mask and developed again. This method is e.g. Currently only possible with negative photoresist. It therefore has the disadvantage that all of the light defects in both photomasks are reproduced, although the effort involved in the two photoresist layers and two masks is accepted. In addition, both light and dark defects are at least partially retained, in that, for example, the thickness of the etching mask formed is reduced by half in each case in their area. For this reason, in addition reasonable in said patent 'encourages conduct to avoid the mask error after the exposure of the first resist any development. Even with this measure, however, light errors cannot be avoided and dark errors can only be partially avoided.

The object on which the invention is based is a photo-etching process due to the random distribution of defects, so-called light and dark defects, in Photomasks caused errors in photolithography in general and in particular in their application in the integrated Circuit technology can be avoided.

According to the invention, this object is achieved in a photo-etching process for the production of masks, in which the surface of a suitable mask material is coated with photoresist, the photoresist exposed and developed via a photomask containing the pattern, and then the etching of the mask material is carried out is achieved in that the photoresist layer several times in succession each time using a separate, but appropriately aligned photomask exposed to a partial exposure that is insufficient for the development of the photoresist layer and that the full exposure is brought about by the sum of the partial exposure.

Docket GE 971 006 309810/0917

2H3737

This method has the particular advantage that the effects all possible mask errors are avoidable, £; o that especially when using the method in the integrated Circuit technology economically justifiable yields can be achieved. An advantageous embodiment consists in that successive photomasks each have at least one have pattern geometry enlarged by the adjustment tolerance. With a slight reduction in the achievable Yield avoided that a perfectly exact mask adjustment is required is.

In particular, the inventive method consists in that at Application of N. £. 3 masks and partial exposures the partial exposures be chosen so that the full exposure is achieved in each case after the (N-1) th partial exposure. An advantageous further development consists in the fact that in the case of a pattern of a photomask composed of regularly arranged individual patterns, only one Extended single mask is used and the partial exposures after moving this mask until the pattern is again covered be made.

The invention is explained in more detail below with reference to two exemplary embodiments shown in the drawing. Show it:

Fig. IA to IB each at the end of individual process steps

the method according to the invention when using two partial exposures,

Fig. IE the obtained using two partial exposures Course of the absorbed light energy,

Fig. IF the corresponding course of the recorded

Light energy if there are only relatively narrow imperfections in the photo masks,

2A to 2E each at the end of individual process steps Docket GE 971 006 309810/0917

2U3737

the method according to the invention when using three partial exposures and

Fig. 2F shows the course of the absorbed light energy

Use of three partial exposures.

As already mentioned, the method according to the invention should in its Application in which the integrated circuit technology are described. The method according to the invention is based on the fact assumed that a complete development of an image pattern on a photoresist layer involves exposure to a fixed requires minimal light energy in relation to the unit area. Usually this minimum light energy is during achieved in a single exposure. It is now made use of that this required minimum light energy also through a series of partial exposures each having a lower light energy can be achieved. The total energy results as the sum of the energies of the partial exposures. Either a separate mask is used for the individual partial exposures the same image pattern or, if the image pattern is composed of regularly arranged individual patterns, the same mask used that after each partial exposure by a single pattern is moved. If one takes into account that the light and dark defects of the photomasks occur in a random distribution, it can thus be assumed that defects in successive photomasks do not come to lie on top of one another. That means, that in the area of a defect there is at most a partial exposure with reduced light energy. Is the light energy one If partial exposure is chosen low enough, the missing steles will be of the photomask in question is not reproduced.

The method according to the invention with only two partial exposures results from FIGS. IA to IE. As an application example serve the production of a diffusion mask in integrated circuit technology. 1A shows a section of a semiconductor substrate S, the surface of which has been oxidized with a

Docket GE 971 006 309810/091 7

I \ HO I O ι

Oxide layer M is covered. The oxide layer M is used at the end of the process as a diffusion mask M ¹ or M ". A photoresist layer PR is applied to the oxide layer M. The photoresist layer PR is in turn covered with a first photomask M1. This photomask shows, for example, light areas 11 and 12 and a dark area 13. Within the dark area, a flaw D1 representing a light defect is assumed. A dark flaw D2 lies within the light area 12. These flaws therefore consist either of the smallest light-permeable spots in parts that are opaque in themselves or of light- Fig. 1A shows the arrangement after the first partial exposure, i.e. below the transparent areas 11, Dl and 12, the photoresist layer PR is partially exposed, which is indicated by the hatched layer parts 14, 16 and 17. There is no underneath the opaque areas 13 and D2 e exposure takes place, which is indicated by the dotted areas 15 and 18. The light defect D1 thus results in an undesirable, partially exposed area 16 in the photoresist layer PR. The dark error D2 has the effect that an area 18 of the photoresist layer PR is unexposed, which area would have to be partially exposed. The light energy supplied during this first partial exposure would not be sufficient for developing the photoresist. It should be pointed out that in the case of a full exposure, instead of this first partial exposure, the layer areas lying below the defects D1 and D2 would either be fully exposed or not hurriedly. These flaws would become real errors in the subsequent development.

In the next step (Fig. IB), the first mask Ml removed and a second, corresponding in the pattern photomask M2 applied. Of course, the second photomask must correspond accordingly the parts that have already been exposed can be precisely adjusted. Since the assumed defects occur only in a random distribution, it can be assumed that the second photomask

OoCcet GE 9H 006 309810/0917

2U3737

M2 the translucent areas 21 and 22 do not have dark defects and the opaque area 23 has no light defect. The only condition is that there are no overlapping defects in the two photomasks. The now carried out second partial exposure causes the · locations of the photoresist layer PR, which were exposed to both exposures, are now fully exposed, which is indicated by the narrow hatched areas 24 and 27 is indicated. The area 18 incorrectly not exposed in the first partial exposure now turns out to be half-exposed area 28 in FIG. 1B. The area 16 incorrectly partially exposed in the first partial exposure becomes not exposed to the second partial exposure, so that in Fig. IB

is retained as a partially exposed area 26. The area 15 not exposed in the first partial exposure is retained as the unexposed area 25. The FIGS. IC and ID show the arrangement after the development of the photoresist, it being necessary to distinguish whether positive or negative photoresist was used. In the arrangement of Fig. IC is believed the use of a negative photoresist PR ^1, which means that the unexposed or only partially exposed areas 25, 26 and 28 disappear, so that openings 25 'and 28' are formed, while the exposed areas 24 and 27 remain as areas 24 '27 and'. Fig. ID shows the arrangement when using a positive photoresist PR ', wherein the unexposed regions 24' and 27 "disappear, ψ while the exposed or partially exposed areas 25", 26 "and 28 ". In the representation chosen here, the limits of the method according to the invention with two partial exposures can be seen. It is evident that only the light error D1 is made to disappear and does not take effect. Defect D2 continues and remains as a defect in the developed photoresist. This is indicated by the areas 28 'in FIG. IC and 28 "in FIG. ID. The fact that when only two partial exposures are used, only one type of defect remains without effect, also results from the course of the light energy recorded, which in FIG. IE is recorded over the width of the overall arrangement.

Docket ge 971 006 309810/0917

is drawn. El corresponds to the energy of the first and E2 the energy of the second partial exposure. The light energy EO corresponds to the minimum amount of energy required for development. The course e21 corresponds to the first partial exposure. The minimum energy is never reached. After the second Partial exposure, the gradient e22 is set. Here, too, it can be seen that in the area of the dark error D2 only one half exposure is achieved, which is insufficient for development, so that the defect remains. From the corresponding The course of the figure in FIG. IF, however, makes it clear that the dark errors D2 also for the most part have no effect when the Defects are relatively small. It should be noted that during the exposure as a result of the scattering effect that occurs a partial exposure also takes place in the first partial exposure below the dark defect D2. This will also apply in the area a small-area dark defect in the second partial exposure reaches the full exposure, so that the defect in the Development disappears. This is shown by the corresponding energy curves e21 'and e22 "in FIG. IF.

If one now looks again at FIGS. IC and ID, it can be seen that the oxide layer M is removed by using a suitable etchant in the areas 25 'or 24 "and 27" from which the photoresist has been removed, and a corresponding diffusion mask M ¹ or M "is retained.

Already when using three partial exposures it is achieved that all possible flaws D1 and D2 will definitely be ineffective. The corresponding process steps and associated Considerations emerge from FIGS. 2A to 2F. The only major difference in the process flow according to the FIGS. 2A to 2D, compared to the process sequence according to FIGS. 1A to ID, consists in the fact that the full exposure is brought about by three partial exposures in conjunction with three separate masks. For the sake of simplicity is again assumed that the defects already in the first mask Ml

Docket _GE 97, 0O ₆ 309810/0917

2H3737

appear. Of course, these fault locations could also be contained in one of the two following masks M2 and M3. It is only essential that, as a result of the random distribution of the flaws, it can be assumed that two flaws of two or even all of the masks never coincide after the alignment of the masks.

The starting point is again a semiconductor substrate S, the surface of which is covered by oxidation with an oxide layer M to be converted ^{into a diffusion mask M 1 or M ″} regions 11 and 12 and an opaque region. 13. in addition, a Es.ll-Feh.le ?: Dl and a dark error D2 "assumed to be present, Bai the first partial exposure with a light energy El (Fig. 2F) _s one third of the total energy at full exposure corresponds to, so the direction indicated by wide hatch partially exposed to one-third portions 14, 16 and 17 are formed _a ünbelichtet remain the regions 15 and 18, 16 and 18. in this case, provide the areas by the voids Dl and D2 caused 2B shows the arrangement after the application of the error-free second mask M2 and implementation of the second Partial exposure. This second photomask is of course aligned with the image pattern that has already been exposed and has the light-permeable areas 21 and 22 and the light-impermeable area 23. During the second partial exposure, two thirds of the areas 14 and 27 are partially exposed (sum of energies E1 and E2 in FIG. 2F). This exposure state is characterized by the denser hatching, the previous area 15, now designated 25, remains unexposed. The defect 26 corresponding to the defect 16 is not affected by the second partial exposure. In contrast, a partial exposure of the defect 18 takes place to a third, which is indicated by the area 28 hatched. In the third partial exposure (Fig. 2C) in conjunction with the third mask .M3,

Docket ge 971 006 309810/0917

2H3737

which has the translucent areas 31 and 32 and the opaque area 33, the areas 24 and 27 from FIG. 2B are fully exposed according to the sum of the three light energies E1, E2 and E3 (FIG. 2F), which is indicated by the narrow hatching of the areas 34 and 37 is indicated. The area 25, now designated by 35, remains unexposed. One third of the defect 16 or 26 caused by the light defect D1 remains partially exposed and is provided with the reference number 36. In this final third partial exposure, it is essential that the area 28 (FIG. 2B) still present in the second partial exposure and partially exposed to one third is now subjected to a further partial exposure. The two-thirds partially exposed area 38 is thereby formed. As can be seen from FIG. 2F, the energies E1, E2 and E3 of the partial exposures are selected so that full exposure is already achieved with two partial exposures. The course e31 of the first partial exposure runs below the minimum energy EO required for full exposure. In the second partial exposure (curve e32), this minimum energy is already exceeded at all desired points. Only in the area of the dark error D2 is the light energy recorded, corresponding to a partial exposure, still below the minimum energy. But this flaw is finally eliminated with the third partial exposure (gradient e33). 2D shows the arrangement when using a negative photoresist PR ¹ after development. The developed photoresist ^{layer PR 1} now forms an etching mask with the etching-preventing areas 34 *, 37 'and 38' and a mask window 35 ¹ , 36 ". Accordingly, a mask window 40 ¹ can be produced ^{in the diffusion mask M 1 by means of an etching process} FIG. 2E shows the production of a complementary arrangement which comes about through the use of a positive photoresist layer PR ″. Here, during the development of the photoresist layer, the unexposed area 35 ″ and the one third partially exposed area 36 ″ are retained, while the fully exposed areas 34 and 37 and the two thirds exposed area 38 are removed

Docket GE 971 006 309 810/0917

2U3737

so that mask windows 34 ″ and 37 ″ or 38 ″ arise in the etching mask formed the area 40 "of the diffusion mask M" is obtained.

It can therefore be seen that the method according to the invention eliminates all possible defects in a photomask, i. H. so hell error and dark defects with negative and positive photoresist, does not have an effect if at least three partial exposures be performed. The partial exposures are to be selected so that the sum of two partial exposures is already gives the exposure required for development, while only a partial exposure is insufficient. How related with the FIGS. IA to IF are shown in the procedure with only two partial exposures under certain circumstances relatively large areas Dark bug not eliminated. But even with the method with only two partial exposures, the distribution is random errors that occur are largely avoided, so that a greater yield can be expected here. this is a Consequence of the light scattering that occurs especially with small imperfections.

For some applications, taking into account the economic viability to simplify the alignment of successive photomasks required in the method according to the invention, the following method can be used with only a slight loss of yield. For example, if you look at the process two partial exposures, the first exposure is made using a regular, the image pattern in the desired dimensions containing photomask performed. In the second exposure, on the other hand, a photomask is used that is at least a the adjustment tolerance has enlarged pattern geometry. Because the image pattern to be transferred is only made up of the sum of both partial exposures is determined in its dimensions, the pattern geometry of the first mask is decisive. Defects in the As a result of the enlarged pattern geometry of the second mask, the area not covered by the two masks will fall in this case

DOOkStGE ₉₇₁ OO ₆ 3098 10/0917

2H3737

Method reproduced and resulted in a loss of yield. All however, further imperfections are still eliminated.

Doolcet _{GE 971} 006 309810/0917

Claims (5)

2H3737 - 14 PATENT CLAIMS Photoetching process in which the surface of a suitable Material coated with photoresist, the photoresist exposed and developed through a photomask containing the pattern and then the material is etched, characterized in that the photoresist layer several times in succession, each time using a separate but containing the same pattern and a correspondingly aligned photomask of a partial exposure which is not sufficient for the development of the photoresist layer is exposed and that the full exposure is brought about by the sum of the partial exposures. 2. Photo-etching process according to claim 1, characterized in that successive photo masks each have at least one have pattern geometry enlarged by the adjustment tolerance. 3. Photo-etching process according to claims 1 and 2, characterized in that that when using η _>. 3 masks and partial exposures the partial exposures are chosen so that after the (n-th partial exposure, the full exposure is achieved. 4. Photo etching process according to claim 3, characterized in that in one of regularly arranged individual patterns composite pattern of a photomask only uses an extended single mask and the partial exposures each time this mask is moved until the pattern is covered again. 5. Application of the photo etching process according to claims 1 to 4, characterized in that it is used for the production of Ln the masks commonly used in semiconductor technology are used. Docket GE 971 006 309810/0917 Empty soap