The invention relates to photolithographic semiconductor fabrication, and more particularly to methods that use masks to transpose (image) circuit structure features onto a semiconductor wafer.
It is known that, in lithography methods, aberrations can occur if the structures to be imaged become very small and have a critical size or a critical distance with respect to one another. The critical size is generally referred to as the “CD” value (CD: Critical Dimension).
What is more, aberrations may occur if structures are arranged so closely next to one another that they influence one another reciprocally during imaging; these aberrations based on “proximity effects” can be reduced by modifying the mask layout beforehand with regard to the “proximity phenomena” that occur. Methods for modifying the mask layout with regard to avoiding proximity effects are referred to by experts by the term OPC methods (OPC: Optical Proximity Correction).
In order to avoid or to reduce these aberrations, it is known to use OPC methods that modify the mask layout 20 beforehand in such a way that the resulting photoresist structure 60 on the wafer 30 corresponds to the greatest possible extent to the desired photoresist structure 25.
In the case of the previously known OPC methods by which a “final” mask layout (c.f. mask 20′ in accordance with
In the case of rule-based OPC methods, the formation of the final mask layout is carried out using rules, in particular tables, defined beforehand. The method disclosed in the two US patents U.S. Pat. No. 5,821,014 and U.S. Pat. No. 5,242,770, by way of example, may be interpreted as a rule-based OPC method, in the case of which optically non-resolvable auxiliary structures are added to the mask layout according to predetermined fixed rules, in order to achieve a better adaptation of the resulting photoresist structure (reference symbol 60 in accordance with
In model-based OPC methods, a lithography simulation method is carried out, in the course of which the exposure operation is simulated. The simulated resulting photoresist structure is compared with the desired photoresist structure, and the mask layout is varied or modified iteratively until a “final” mask layout is present, which achieves an optimum correspondence between the simulated photoresist structure and the desired photoresist structure. The lithography simulation is carried out with the aid of a, for example, DP-based lithography simulator that is based on a simulation model for the lithography process. For this purpose, the simulation model is determined beforehand by “fitting” or adapting model parameters to experimental data. The model parameters may be determined for example by evaluation of so-called OPC curves for various CD values or structure types. One example of an OPC curve is shown in
Irrespective of whether a model-based or a rule-based OPC method is involved in an OPC method, OPC variants can also differ from one another with regard to their respective optimization aim. For example, so-called “target” OPC method and so-called process window OPC methods, for example “defocus” OPC methods, have different optimization aims:
The aim of target OPC methods is to hit as accurately as possible the prescribed target for the individual geometric dimensions of the mask structures in the case of correctly observing all the prescribed technological and methodological conditions (for example focus, exposure to light, etc.). Thus, in the case of a target OPC variant it is assumed that all prescribed process parameters are “hit” or set and observed in an ideal way. In this case the term “target” is understood to mean the structural size of the main structures to be imaged.
Since the gate length of transistors is of decisive importance for their electrical behavior, target OPC methods are used, in particular, for the gate level of masks. However, it is disadvantageous in the case of the target OPC variant that the prescribed geometric dimensions of the mask structures are actually observed only when the prescribed process parameters are observed in a quasi-exact fashion. If fluctuations arise in the process parameters, it is possible for sometimes considerable deviations to occur between the desired mask structures or mask dimensions and the actually resulting mask structures and mask dimensions; this can lead, for example, to breaking of lines or to a short circuit between lines. The resulting process window is therefore generally relatively small in the case of a target OPC method.
By contrast, process window OPC methods, for example defocus OPC methods, have the aim of rendering as large as possible the process window—that is to say the permissible parameter range of the process parameters for the exposure process with the resulting mask—in order to ensure the observation of the mask specifications even in the case of process fluctuations. It is accepted in this case in defocus OPC methods that the target geometric mask dimensions are not met exactly; deviations are therefore deliberately accepted in order to enlarge the process window and thus the tolerance range during later use of the mask.
A defocus OPC method is described, for example in the above named German patent DE 101 33 127. In this method, a “fictitious” defocus value is prescribed which forms a basis for simulating the exposure operation; this defocus value specifies that the resist structure to be exposed with the mask is situated somewhat outside the optimal focal plane. Despite the supposedly present defocusing, in the course of the OPC method an attempt is made to achieve an optimum imaging behavior of the mask; thus, an attempt is made to compensate the aberration caused by the supposed defocusing. The effect of this “compensation operation” is to change the form of the mask layout in such a way that both the line structures are formed more widely, and that a larger distance is produced between two adjacent line structures in each case. A mask is therefore obtained with the aid of which, given the use of a focused exposure, the probability of the formation of wider line structures, and the formation of larger distances between adjacent line structures in each case is higher than the probability of the formation of excessively small line structures and the formation of excessively small distances between adjacent line structures.
The invention is directed to an improved method of the type specified in the introduction to the effect that aberrations, in particular as a result of proximity effects, are reduced even better than before.
Accordingly, it is provided according to the invention that in the course of the OPC method at least two different OPC variants are used subdividing the original auxiliary mask layout into at least two layout areas and processing each of the layout areas in accordance with one of the at least two OPC variants.
A substantial advantage of the method according to the invention consists in that each of the layout areas is optimized with an individually assigned OPC variant that is particularly suitable in each case for the respective layout area. According to the invention therefore, there is no “overall”, mask-embracing identical optimization for all layout areas of the layout, but instead an individual optimization referred to layout areas. The result of this mode of procedure is that after termination of the OPC method a final mask layout is present with which a particularly high process stability is achieved. The term “process stability” is understood here to mean that a sufficiently large process window is achieved, on the one hand, and that prescribed mask parameters or target parameters are achieved optimally, on the other hand.
A further substantial advantage of the method according to the invention consists in that fewer OPC process cycles or OPC passes are required overall until the optimum, final mask layout has been determined than is the case with the previously known “pure” target OPC and “pure” defocus OPC methods. A substantial process acceleration is thus additionally achieved because of the separation of the mask layout into at least two layout areas, and of the optimization of these layout areas for each of them individually.
A third substantial advantage of the method according to the invention consists in that there is generally no need for post-processing. Post-processing is understood here to mean that mask defects possibly present in the final mask layout are removed manually or with further optimization programs using DV systems. Specifically, because the optimization is in terms of individual layout areas there is in general no need in the method according to the invention for manual “reworking” of the mask layout, because mask defects occur only seldom enough to be negligible.
A fourth substantial advantage of the method according to the invention consists in that it is possible to appeal to already known and tried and tested OPC variants. It is necessary merely to carry out a subdivision of layout areas and to select the OPC variant particularly suitable in each case for the respective layout areas of the mask layout.
Since—as stated above—process window OPC methods, in particular defocus OPC methods, and target OPC methods have already been tested in practice, it is considered as advantageous within the course of developing the method if the method comprises at least one process window OPC variant, in particular a defocus OPC variant, and/or at least one target OPC variant.
As has already been explained at the beginning, gate structures of transistors, in particular, are especially critical, since with these structures it is particularly important to observe the prescribed geometric dimensions, in particular the gate length. It is therefore considered as advantageous within the course of a further improvement of the method according to the invention if the provisional auxiliary mask layout is subdivided into a layout area with active structures and into a layout area with inactive structures such that it is possible to go into the specific optimization requirements of the active structures, in particular.
It is preferred for the layout area with the active structures to be subjected to the target OPC variant, and for the layout area with the inactive structures to be subjected to the defocus OPC variant. In this case, the gate structures of transistors are preferably treated as the active structures.
The active layout areas can be determined particularly simply and therefore advantageously when the provisional auxiliary mask layout and the mask layout, describing the active areas and thus the gate structures, of masks prearranged in terms of layout are “laid over one another” by software or by hand, and those areas which are situated over active zones—for example diffusion zones —are treated as active structures. The masks which define the diffusion zones, for example, can be used for the “process of laying one over another”.
It is also considered advantageous when transition areas (“buffer zones”) are formed between the active and the inactive layout areas, and when these transition zones are optimized separately; the buffer zones can also be assigned to the target OPC variant, for example.
Both for the defocus OPC variant and for the target OPC variant, it is advantageously possible to select in each case either a model-based OPC variant or a rule-based OPC variant; however, model-based variants are to be preferred in each case.
It is also considered as advantageous when a modified auxiliary mask layout is firstly formed before carrying out the OPC method with the provisional auxiliary mask layout by supplementing the mask structures of the provisional auxiliary mask layout in a first modification step with the formation of modified mask structures in accordance with prescribed placing rules by means of optically non-resolvable auxiliary structures with the formation of the modified auxiliary mask layout, and the final mask layout is produced with the aid of the OPC method by using the modified auxiliary mask layout. Reference may be made to the U.S. Pat. Nos. 5,821,014 and 5,242,770 with reference to the mode of procedure when adding optically non-resolvable auxiliary structures.
It is also considered as advantageous when those layout areas which are provided with the optically non-resolvable auxiliary structures are subjected to a different OPC variant from the layout areas without the optically non-resolvable auxiliary structures. For example, the layout areas with the optically non-resolvable auxiliary structures can be subjected to a target OPC variant, and the surrounding layout areas are subjected to a defocus OPC variant.
Since the layout areas with the active structures are very critical, because the electrical behavior of the later electric components is generally a function of these structures, it is considered to be advantageous when firstly the layout areas with the active structures, in particular firstly the layout areas with the gate structures, are supplemented by means of optically non-resolvable auxiliary structures. Only when the active structures with the optically non-resolvable auxiliary structures are provided it is then also possible to optimize the remaining layout areas appropriately.
It is also considered to be advantageous when only the layout areas with the active structures, in particular only the layout areas with the gate structures, are supplemented by means of the optically resolvable auxiliary structures. To be precise, it is thereby excluded that optically non-resolvable auxiliary structures for inactive layout areas can impair adjacent active layout areas.
Moreover, it is preferably ensured when producing the final mask layout that wiring areas are not situated over active layout areas if no contact is to remain with the latter.
Moreover, when producing the final mask layout, pad structures (for example “landing pads”) and remaining wiring structures are preferably treated differently, since because of the subsequently required contacting steps pad structures are subject to different requirements than the remaining wire structures. It is preferred to use a target OPC variant when producing the mask layout for the pad structures.
It is also considered as advantageous when the method is carried out—in particular only—on non-gate levels, in particular on RX areas (=diffusion zones in the case of logic chips) and/or on metallization levels.
Moreover, CD critical and non-CD critical structures can advantageously be treated with different OPC variants.
The method can be used, for example, for DRAM mask layouts. It is then preferred in this case to process cell field structures and cell field edge structures with different OPC variants; this is because the cell field edge frequently consists of dummy structures which are not required functionally in electrical terms. Thus, electrically necessary mask structures and unnecessary dummy structures are preferably treated differently.
In order to elucidate the invention,
In this case, the OPC curve 70 describes the CD value profile on the wafer given a constant mask CD value, which is likewise depicted in
The photoresist structure 60 defines interconnect structures 110 in accordance with which interconnects are produced in subsequent process steps, for example by etching or by vapor deposition processes using the photoresist structure 60. The interconnect structures 110 thus image a metallization level, a gate contact connection and a wiring structure for one or more electronic components which are monolithically integrated in the silicon substrate 100.
Also to be seen in
The diffusion zones 120 can therefore also be denoted as “active” zones; the remaining zones of silicon substrate 100, which are situated outside these active zones, are denoted below as “passive” zones of silicon substrate 100. Passive zones are used, for example, for providing the active elements or as landing pads (contact hole landing points).
Those interconnect structures which are situated over the active zones of the silicon substrate 100, and therefore over the diffusion zones 120, and make contact therewith are denoted below as active structures 130. The remaining interconnect structures are subsequently known as inactive or passive structures 140 in what follows.
The mask layout 20′ therefore has “active” layout areas 130′ with the active structures 130, and “passive” layout areas 140′ with the inactive structures 140.
Also to be seen in
Visible in
In addition to the interconnect structures 100,
In order to clearly display the position of the SRAF structures 170 relative to the interconnect structures 110 and to the diffusion zones 120,
It is to be seen in
The method according to the invention will now be explained with the aid of a first exemplary embodiment in conjunction with
Visible in
The provisional auxiliary mask layout 200 has active layout areas 230′ and inactive layout areas 240′. As already explained above, the active layout areas 230′ are understood here to be those layout areas of the provisional auxiliary mask layout 200 which are arranged over the active zones of the electric circuit. Active zones are formed, for example, by the diffusion zones of transistors with which contact is made by the active layout areas 230′ of the auxiliary mask layout 200.
As above, inactive layout areas 240′ are understood to be those layout areas which are arranged outside active zones and therefore essentially form connections between the active layout areas 230′ or to external landing pads (or contact hole landing points) 250.
Before carrying out an OPC method, the provisional auxiliary mask layout 200 is subdivided in a first preparatory step into the active layout areas 230′ and into the passive or inactive layout areas 240′ SO that these can be treated differently.
In order to define which layout areas of the preliminary auxiliary mask layout 200 are active and which are the inactive layout areas, the provisional auxiliary mask layout 200 is laid over the mask layout of the prearranged masks—that is to say over the mask layout of those masks which are processed in front of the mask with the mask layout 200 in accordance with
Those layout areas of the provisional auxiliary mask layout 200 which are situated over active zones (transistor zones, in particular diffusion zones or gate areas of transistors) are therefore identified as active layout areas 230′; the remaining layout areas are identified as inactive layout areas 240′.
The provisional auxiliary mask layout 200 is then used in the course of an OPC method to form a final mask layout 300;
When carrying out the OPC method, the active layout areas 230′ and the passive layout areas 240′ are treated differently—by contrast with the OPC method of the prior art—since the active layout areas and the passive layout areas are subject to a different tolerance requirement. In concrete terms, layout areas 230′ with the active structures 230 are subjected to a target OPC method (or a target OPC variant), whereas the layout areas 240′ with the passive structures 240 are optimized using a defocus OPC method (or defocus OPC variant).
In the course of the OPC method, optically non-resolvable SRAF auxiliary structures 350 are optically added to the interconnect structures 210. The purpose of these SRAF structures is to improve the imaging behavior of the mask layout. Alternatively, the SRAF structures can also be added in a rule-based fashion before carrying out the OPC method; in such a case, a modified auxiliary mask layout is firstly formed from the auxiliary mask layout 200 and the OPC method is carried out using said modified auxiliary mask layout.
It is, however, possible to form in the boundary areas between the active layout areas 230′ and the inactive layout areas 240′ buffer zones 360 which are optionally treated like the active layout areas 230′, like the passive layout areas 240′ or in accordance with a dedicated optimization method. The buffer zones 360 serve to take account of, and as far as possible to compensate, overlay fluctuations or the offsetting of different lithography levels from one another.
As may be seen from figure. 6, it is possible in the case of conventional placing of the optically non-resolvable auxiliary structures 350 to detect areas in which the optically non-resolvable auxiliary structures (SRAFs) 350 are not arranged optimally. For example, optically non-resolvable auxiliary structures 350 are set in a non-parallel fashion along the gate (active structure) in area 370, it thereby being possible for aberrations to occur in the case of the CD-critical active structure. The wiring structure (passive structure) located to the right above the gate is, by contrast, erroneously optimally supported by SRAFs 350. The “gate first approach” described further below in conjunction with
Because of the excessively small distance between the wiring interconnects, no SRAF structures 350 are arranged in another area 380 of
The placing of the optically non-resolvable auxiliary structures 350 is undertaken in another way in accordance with a second exemplary embodiment of the method according to the invention. This is shown in accordance with
It is to be seen that in the exemplary embodiment in accordance with
Placing the optically non-resolvable auxiliary structures 350 in the region of the active layout areas 230′ can also be denoted as “gate first replacement”, since the optically active areas 230′ are formed by the gate regions of transistors.
In a third exemplary embodiment of the method, the SRAF auxiliary structures 350 are arranged exclusively in the region of the active layout areas 230′ (compare
It is to be seen in
It is also to be seen in
Just like the active layout areas, landing pads can likewise be detected from mask layouts being laid one on top of the other; alternatively, the landing pads can also be detected manually or with the aid of a data processing system by using their typical geometric dimensions or by using the specifications of the electric circuit.
Number | Date | Country | Kind |
---|---|---|---|
103 56 693.7 | Nov 2003 | DE | national |