JP4018309B2 - Circuit parameter extraction method, semiconductor integrated circuit design method and apparatus - Google Patents

Description

[0001]
[Technical field to which the invention belongs]
The present invention mainly relates to a method for extracting circuit parameters such as a wiring resistance value, a wiring capacitance value, and a gate width of a transistor in semiconductor integrated circuit design, and a semiconductor integrated circuit for executing a timing verification simulation of a semiconductor integrated circuit using the extracted circuit parameters. The present invention relates to a circuit design method and apparatus.
[0002]
[Prior art]
In conventional semiconductor integrated circuit design, when circuit data for simulation for timing verification is extracted / created from layout data, wiring capacitance is calculated in consideration of the relationship with surrounding wiring. The wiring resistance was calculated considering only the width and length of the wiring itself drawn like a mask. At least, when performing a timing simulation used in combination with an automatic placement and routing tool currently on the market, the wiring interval with the same layer wiring is not considered.
[0003]
However, the fact that the wiring resistance has no relation to the surrounding wiring and the wiring capacity is determined by the relation with the surrounding wiring is correct from a physical phenomenon point of view, but in the actual semiconductor manufacturing process, the wiring pattern formation is Since it is formed by the lithography process or etching process, the finished wiring width or cross-sectional shape depends on whether the wiring of the same layer as the target wiring exists around, the distance to the surrounding wiring, the area ratio of the wiring pattern, etc. The difference comes out.
[0004]
In particular, in recent years, process miniaturization has been accelerated, and the margin of the wavelength of the stepper light source used in the lithography process is gradually decreasing with respect to the minimum dimension of the pattern to be processed in the lithography process. For this reason, optical proximity correction (Optical Proximity Correction, abbreviated as OPC) is devised by detecting the distance between the wiring patterns and applying a correction to the mask pattern so as to obtain a desired finished width. This OPC is generally used in recent fine processes.
[0005]
With regard to the metal wiring layer, the point that the lithography process is used for pattern drawing is the same as that of the gate electrode layer. However, in the same process generation, the minimum processing dimension is generally larger than that of the gate electrode layer. More specifically, in the 0.18 μm generation, the general layout rule is that the gate electrode width is about 0.18 μm and the metal wiring width is about 0.20 to 0.35 μm. Since metal wiring has not been required to have a strict standard value that is more severe than the gate dimension that directly affects the transistor characteristics, OPC or the like is generally rarely used at present. In other words, desired resist dimension control (lithographic performance) can be ensured without using OPC.
[0006]
In addition, the formation of the isolation region of the diffusion layer is also shifting from the conventional LOCOS to STI (Shallow Trench Isolation) with the miniaturization of the process. In LOCOS, SiO for separation by thermal oxidation of silicon. ₂ Although a film was formed, it had a disadvantage that a large amount of penetration called a bird's beak occurred and was not suitable for miniaturization. The STI has been devised to avoid the problem of bird's beak generation and form a fine transistor isolation region. As an outline, SiO etched on the part carved by dry etching ₂ It is a process flow for forming a separation membrane by a flow for embedding a membrane. In addition to the advantage that a fine separation pattern can be formed, LOCOS has a large parasitic capacitance component called channel stopper implantation on the side surface of the diffusion layer, whereas STI has SiO. ₂ Since the side surface of the transistor active region is separated by the film, there is an advantage that the parasitic capacitance is small.
[0007]
[Problems to be solved by the invention]
However, the conventional method has the following problems.
[0008]
It should be considered that the wiring manufacturing process includes not only a wiring lithography process but also a wiring dry etching process after resist formation. Hereinafter, the LSI manufacturing process will be described with reference to FIG.
[0009]
As shown in FIG. 13, the wiring manufacturing process includes (a) depositing a wiring layer, (b) applying a resist, (c) exposing using a mask, (d) removing unnecessary resist patterns, and (e) resist. The wiring layer is patterned by dry etching using a pattern.
[0010]
First, a resist pattern is formed in steps (c) and (d), but due to the optical interference with the pattern existing in the vicinity and the area ratio, it is not finished according to the pattern size drawn on the mask. There is. For this problem, the optical proximity correction (OPC) used for the gate electrode as shown in the prior art is effective.
[0011]
Furthermore, even if the resist is formed according to the dimensions in the lithography process, there is a difference in the wiring shape depending on whether or not another wiring exists near the wiring pattern in the dry etching process in the wiring pattern formation. There is a new problem.
[0012]
It is assumed that the wiring etching conditions are optimally set to the minimum line & space pitch or a narrow pitch that is used very frequently, such as a wiring grid in an automatic placement and routing tool in the ASIC design method. Here, the optimum setting means that the etching conditions are set so that the wiring cross-sectional shape to be etched becomes a substantially square in such a layout state.
[0013]
In such an etching condition setting, for a layout pattern in which no other wiring exists around a certain model wiring, the bottom portion of the model wiring tends to be thick and the wiring resistance value tends to be low. The reason for this is that when there is no wiring around, the amount of metal to be removed during dry etching becomes very large, and the metal to be etched once deposited by plasma etching is deposited again on the side walls of the wiring pattern, etc. It is explained by. (See the cross-sectional view of the line and space pattern, the cross-sectional view of the isolated pattern, and the enlarged view of FIG. 14 (b) shown in FIGS. 14 (a), 14 (b), and 14 (c)), respectively.
Therefore, the farther away the wiring next to the target wiring is, the more metal is blown near the side wall during etching. There is a problem that the lower base becomes larger trapezoidal shape than the upper base where the width of the part is widened.
[0014]
FIG. 15 shows a dependency graph (measured value) between the wiring space and the finished wiring width. As can be seen from FIG. 15, the dimension of the top portion of the wiring hardly changes even when the distance between the wirings (space width) changes, and the dimension on the bottom side changes. From this, it can be seen that no matter how much the optical proximity effect correction or the isolated pattern correction is performed so that the resist dimensions are properly obtained by taking care of the lithography, the thickness of the bottom portion cannot be accurately handled. .
[0015]
More precisely, the above explanation was given for the reason that the etching plasma flies perpendicularly to the wiring pattern from the top of the resist so that it cannot be cut beyond the resist dimensions. Etching plasma having a velocity component that jumps into the side surface of the wiring may cause the wiring top portion dimension under the resist to be cut beyond the resist dimension.
[0016]
In addition, the pattern area ratio factor is an important factor for such process steps, but resist lithography, which has been discussed using the optical proximity effect and the like, has been conventionally used for this wiring dry etch step. Also for the process, the area ratio of the target pattern is not taken into account, and by taking it into account, the finished pattern shape of the wiring, and circuit parameters such as wiring resistance and capacitance used in the function confirmation simulation derived therefrom It was found that the accuracy could be further improved.
[0017]
FIG. 16 shows correlation data (actually measured values) between the wiring pattern area ratio (opening ratio = 100−pattern area ratio) and the amount of deviation of the wiring finished width in the process of scraping the metal in the dry etching process. As shown in FIG. 16, it can be seen that the smaller the wiring pattern area ratio (the larger the aperture ratio), the thicker the finished wiring width. Incidentally, since the gate electrode is made of a material that is etched differently from that of the wiring, the taper angle applied during the etching is not as problematic as the wiring.
[0018]
Recently, a process using copper (Cu) instead of aluminum as a wiring material has been used for the purpose of reducing the resistance / high reliability of the wiring. In the case of Cu, etching is difficult, and the current mainstream process is to use a damascene process in which an interlayer film is dug, Cu is embedded therein, and Cu is formed by CMP (Chemical Mechanical Polishing). It is. However, since the etching process is also used in this damascene process, the problem that the taper angle changes depending on the distance between the wires and the area ratio is the same. However, the relationship between the positive and negative is opposite to the etching of the wiring itself, and the correlation between the distance between wirings, the wiring pattern area ratio and the finished width is the reverse of the increase and decrease compared to the etching of the wiring itself. . Further, in a pattern having a large wiring width, a defect called dishing, in which the wiring width becomes thinner in a mortar shape at the time of metal CMP, is different from the process of etching the wiring itself.
[0019]
Also, in the above-described transistor isolation region formation process STI, an etching process exists when the isolation region is dug, and the details depend on the etching characteristics, but if the distance between the transistor active regions changes, the STI end There is a possibility that the taper angle changes and the transistor gate width changes.
[0020]
Therefore, the present invention has been made in view of the above problems, and a circuit parameter extraction method that enables highly accurate wiring finish width calculation and highly accurate circuit simulation, a method for designing a semiconductor integrated circuit using the circuit parameter extraction method, and An object is to provide an apparatus.
[0021]
[Means for Solving the Problems]
In order to achieve the above object, a first circuit parameter extracting method of the present invention is a method for extracting circuit parameters such as wiring resistance and wiring capacitance from a layout of a semiconductor integrated circuit, and includes a model wiring and a periphery of the model wiring. In addition to preparing the correlation data between the distance between the wiring of the same layer and the difference between the mask layout width and the finished width of the model wiring, and extracting the wiring length and wiring width of the analysis wiring from the actual layout The distance between the analysis wiring and the wiring existing in the same layer is extracted, the extracted layout wiring width of the analysis wiring, and the analysis wiring and the wiring existing around the analysis wiring are also extracted. A wiring resistance value and a wiring capacitance value are calculated using a wiring finish width obtained by referring to the correlation data with respect to the distance.
[0022]
According to this first circuit parameter extraction method, by using correlation data between the finished width of the analysis wiring, which has not been considered in the prior art, and the distance between the analysis wiring and the wiring existing around the analysis wiring, the first circuit parameter extraction method can be used. High-precision wiring resistance / capacitance extraction and high-precision circuit simulation are possible. As a result, it is possible to suppress timing errors such as timing skew that may be caused by a difference in wiring resistance value caused by a difference in wiring distance at each location on the chip.
[0023]
In addition, since this method can be used including the change in the wiring finish width caused by the lithography process, even if the wiring top portion dimension is not finished as the mask layout pattern, the difference is also possible. In addition, wiring parameters can be extracted and simulated with high accuracy. As an application method, the resist used for wiring is OK level that does not lead to disconnection or short-circuit according to the processing rule, but the standard value for the finish variation is required to have an accuracy that cannot be realized without using OPC. In such a case, by using the circuit parameter extraction considering the inter-wiring distance and the wiring finish width dependency of the present invention without OPC, a high-accuracy high-priced mask that can handle fine additional patterns for OPC and a long-time OPC mask No processing is required.
[0024]
In order to achieve the above object, a second circuit parameter extraction method of the present invention is a method for extracting circuit parameters such as wiring resistance and wiring capacitance from a layout of a semiconductor integrated circuit, and is provided in a wiring layer of a model wiring. Prepare the correlation data between the wiring pattern area ratio, which is the ratio of wiring patterns existing in the setting area, and the amount of change in the finished width of the model wiring, and extract the wiring length and wiring width of the analysis wiring from the actual layout. And calculating the pattern area ratio of the wiring layer of the analysis wiring, and refer to the correlation data for the extracted wiring length and wiring width of the analysis wiring and the calculated pattern area ratio of the analysis wiring layer. Using the amount of change in the wiring finish width due to the area ratio obtained by performing the above, the wiring finish width is obtained, and the wiring resistance value and wiring capacitance value are calculated. That.
[0025]
According to the second circuit parameter extraction method, it is possible to extract wiring resistance and wiring capacitance parameters with high accuracy by using correlation data between the wiring layer area ratio of the analysis wiring and the finished width of the analysis wiring. In addition, it is possible to suppress timing errors such as timing skew that may be caused by differences in wiring resistance values caused by wiring density differences or wiring distance differences at various locations on the chip.
[0026]
In the first and second circuit parameter extraction methods , Associate The correlation data to be provided Is The cross section of the model wiring has a rectangle, a quadrilateral having an acute angle or an obtuse angle other than a rectangle, a polygon of four or more squares, a shape expressing a shape of a side as a circumferential curve with a certain curvature, and a certain side. Expressed as a change in the shape expressed by a function curve, in addition to the extracted wiring length and wiring width on the analysis wiring layout, the finished wiring cross-sectional shape obtained by referring to the correlation data It is preferable that the wiring resistance value per unit length is obtained by calculating the cross-sectional area of the wiring, and the resistance value and capacitance value of the analysis wiring are calculated from the wiring resistance value per unit length and the wiring length on the layout.
[0027]
According to this method, it is possible to perform resistance calculation with higher accuracy than the conventional one treated as a square section.
[0028]
The wiring cross-sectional shape is preferably handled as a trapezoidal shape.
[0029]
According to this method, it is possible to perform resistance calculation with higher accuracy than the conventional one treated as a square section. In particular, the trapezoidal shape is not so complicated to handle, and is an approximate shape that can handle the finished wiring shape with considerably high accuracy.
[0030]
In order to achieve the above object, a third circuit parameter extraction method of the present invention is a method for extracting circuit parameters such as wiring resistance and wiring capacitance from a layout of a semiconductor integrated circuit, and includes a wiring width on a model wiring layout. Is used as a parameter to prepare correlation data expressing the distance between the model wiring and the wiring in the same layer existing around the model wiring and the amount of change in the resistance value and capacitance value of the model wiring, and the actual layout From this, the wiring length and the wiring width of the analysis wiring are extracted, the distance between the analysis wiring and the wiring existing in the same layer is extracted, and the correlation data is extracted using the extracted layout wiring width of the analysis wiring as a parameter. The finish resistance value and capacitance value of the analysis wiring are calculated by referring to FIG.
[0031]
According to the third circuit parameter extraction method, in the case where the layout with a determined wiring width is the main as in the ASIC design method, it is more efficient to use the resistance value of the wiring of interest directly with less calculation amount. The circuit simulation can be performed in consideration of the influence of the distance between adjacent wirings on the wiring resistance.
[0032]
In order to achieve the above object, a fourth circuit parameter extracting method of the present invention is a method for extracting circuit parameters such as wiring resistance and wiring capacitance from a layout of a semiconductor integrated circuit, and includes a wiring width on a layout of a model wiring. Is used as a parameter to prepare correlation data between the wiring pattern area ratio, which is the proportion of the wiring pattern existing in the set area where the wiring layer of the model wiring is, and the amount of change in the resistance value and capacitance value of the model wiring. The wiring length and wiring width of the analysis wiring are extracted from the layout, and the pattern area ratio of the wiring layer of the analysis wiring is calculated, and the correlation data is referred to using the extracted layout wiring width of the analysis wiring as a parameter. Thus, the finished resistance value and capacitance value of the analysis wiring are calculated.
[0033]
According to the fourth circuit parameter extraction method, in the case where the layout with the determined wiring width is the main as in the ASIC design method, it is more efficient to use the resistance value of the wiring of interest directly with less calculation amount. In addition, it is possible to perform circuit simulation considering the influence of the wiring layer area ratio on the wiring resistance.
[0034]
In the first to fourth circuit parameter extraction methods, , Associate The correlation data to be provided Is , Table or approximate formula Has been It is preferable.
[0035]
According to this method, in the table expression, simulation can be performed with the desired value itself prepared on the basis of the actual measurement value at the table point (interpolation between points is usually performed by a linear function or higher order function) and approximate calculation In terms of expression, high-precision fitting can be performed at successive points related to wiring space and the like.
[0036]
In the second and fourth circuit parameter extraction methods, it is preferable to calculate the area ratio of the wiring pattern in the entire chip area when calculating the wiring pattern area ratio.
[0037]
According to this method, as long as there is no significant density of wiring patterns in the chip, it is possible to calculate wiring resistance and wiring capacity with high accuracy using the wiring pattern area ratio with a small number of CAD processes and CAD operations.
[0038]
Alternatively, in the second and fourth circuit parameter extraction methods, it is preferable to calculate the wiring pattern area ratio for each circuit block when calculating the wiring pattern area ratio.
[0039]
According to this method, when the area ratio is calculated, the area ratio with respect to the block area is used, so that the area ratio can be calculated several times to several tens of times. Therefore, a high-load CAD process is not performed. In addition, circuit parameters can be extracted in consideration of the wiring area ratio. In addition, at the block level, the characteristics of whether the number of wirings is large or small, whether a large number of transistors having a large gate width are used, and which of the transistors and wirings determines the area are often determined. It becomes a very effective method when the amount is taken into consideration.
[0040]
Alternatively, in the second and fourth circuit parameter extraction methods, when the wiring pattern area ratio is calculated, the patterning process is performed within a range in which the wiring processing process can be affected by the area ratio of the wiring pattern existing around. It is preferable to calculate the area ratio.
[0041]
According to this method, when the area ratio is calculated, the area ratio is calculated by taking the area where the target process step is affected. Accurate circuit parameter extraction is possible.
[0042]
Moreover, it is preferable to calculate the wiring pattern area ratio within a region range 10 times the minimum processing dimension.
[0043]
As a result, when the process with a large dependency on the wiring layer area ratio is the wiring etching process, the area where the wiring area ratio greatly affects the amount of change in the wiring finished width with respect to the wiring layout width Therefore, the circuit parameters can be extracted with high accuracy.
[0044]
In order to achieve the above object, the first design method of a semiconductor integrated circuit according to the present invention is caused by the first to fourth circuit parameter extraction methods and the lithography process in the photomask data generation process used in the semiconductor lithography process. The optical proximity effect correction for correcting the amount of change in the resist finish width and the correction for the amount of change in resist thinning in an isolated pattern caused by the resist etching process are used in combination.
[0045]
According to this first design method, the optical proximity effect correction in the semiconductor lithography process and the resist thinning correction of the isolated pattern can be corrected, and the deviation from the finished shape caused by the wiring etching can be corrected. It becomes possible.
[0046]
In order to achieve the above object, a second design method of a semiconductor integrated circuit according to the present invention refers to a mask layout data and uses the first to fourth circuit parameter extraction methods to obtain a wiring finish width and a wiring resistance. After calculating the value and the wiring capacitance value, the wiring resistance value and the wiring capacitance value are compared with the desired wiring resistance value and the wiring capacitance value set at the time of design, and the desired wiring resistance value is compared with the layout wiring width. And determining whether or not correction to approximate the wiring capacitance value is to be applied. If correction is necessary as a result of the determination, the layout wiring width is corrected to be close to the desired wiring resistance value and wiring capacitance value. In addition, the timing simulation simulation of the semiconductor integrated circuit is performed again using the wiring resistance value and wiring capacitance value after correcting the layout wiring width.
[0047]
According to the second design method, not only can the resistance value and the capacitance value at the time of the finish of the wiring to be analyzed be accurately handled, but also a process to bring it closer to the resistance value assumed at the time of logic & timing design is performed. The verification result in the final timing verification simulation is more likely to be OK. In addition, it is possible to suppress timing errors such as timing skew that may be caused by differences in wiring resistance values caused by wiring density differences or wiring distance differences at various locations on the chip.
[0048]
In order to achieve the above object, a third design method of a semiconductor integrated circuit according to the present invention treats the cross-sectional shape of a wiring as a trapezoidal shape, and corrects an optical proximity effect correction for correcting a change amount of a resist finish width caused by a lithography process. And layout correction for correcting the amount of change in resist thinning in an isolated pattern caused by the resist etching process, and using the layout correction, the model wiring and the wiring in the same layer existing around the model wiring Together with the first correlation data of the top and bottom of the finished shape of the model wiring, the area ratio of the wiring layer of the model wiring using the layout correction, and the model wiring Prepare the second correlation data with the finished values of the top and bottom of the finished shape, and extract the wiring length and wiring width of the analysis wiring from the actual layout The distance between the analysis wiring and the wiring existing in the same layer is extracted, the pattern area ratio of the wiring layer of the analysis wiring is calculated, and the wiring length, the wiring width, and the analysis wiring of the extracted analysis wiring are calculated. Using the distance to the wiring existing in the same layer and the pattern area ratio of the wiring layer of the analysis wiring, the amount of change in the upper and lower bases in the trapezoidal cross section of the analysis wiring, A wiring cross-sectional area and a wiring resistance value are calculated, and it is determined whether or not to correct the calculated wiring resistance value to be approximated to a desired wiring resistance value and wiring capacitance value intended at the time of design. As a result, when correction is necessary, correction is made to change the layout wiring width so as to approach the desired wiring resistance value and wiring capacitance value, and the wiring resistance value and wiring capacitance after the correction is performed again. Use value , And performs the timing verification simulation of the semiconductor integrated circuit.
[0049]
According to the third design method, the first correlation data between the finished width of the analysis wiring, which has not been considered in the prior art, and the distance between the analysis wiring and the wiring existing around the analysis wiring, and the wiring layer area ratio of the analysis wiring And the second correlation data between the analysis wiring finish width are handled as a trapezoidal cross-sectional shape that is relatively easy to handle and has high accuracy, so that wiring resistance and wiring capacity such as finish can be easily and accurately extracted. & By performing the process of approaching the resistance value assumed at the time of timing design, the verification result in the final timing verification simulation is more likely to be OK.
[0050]
In the third design method, The wiring layer of the analysis wiring Pattern area ratio Calculate At this time, it is preferable to calculate the area ratio of the wiring pattern in the entire chip area.
[0051]
According to this method, as long as there is no significant density of wiring patterns in the chip, it is possible to calculate wiring resistance and wiring capacity with high accuracy using the wiring pattern area ratio with a small number of CAD processes and CAD operations.
[0052]
Alternatively, in the third design method, The wiring layer of the analysis wiring Pattern area ratio Calculate At this time, it is preferable to calculate the area ratio of the wiring pattern for each circuit block.
[0053]
According to this method, when the area ratio is calculated, the area ratio with respect to the block area is used, so that the area ratio can be calculated several times to several tens of times. Therefore, a high-load CAD process is not performed. In addition, circuit parameters can be extracted in consideration of the wiring area ratio. In addition, at the block level, the characteristics of whether the number of wirings is large or small, whether a large number of transistors having a large gate width are used, and which of the transistors and wirings determines the area are often determined. It becomes a very effective method when the amount is taken into consideration.
[0054]
Alternatively, in the third design method, The wiring layer of the analysis wiring Pattern area ratio Calculate At this time, it is preferable to calculate the pattern area ratio within a range in which the wiring processing process step can be influenced by the area ratio of the wiring pattern existing around.
[0055]
According to this method, when the area ratio is calculated, the area ratio is calculated by taking the area where the target process step is affected. Accurate circuit parameter extraction is possible.
[0056]
In addition, in the region range of 10 times the minimum processing dimension, The wiring layer of the analysis wiring It is preferable to calculate the pattern area ratio.
[0057]
As a result, when the process with a large dependency on the wiring layer area ratio is the wiring etching process, the area where the wiring area ratio greatly affects the amount of change in the wiring finished width with respect to the wiring layout width Therefore, the circuit parameters can be extracted with high accuracy.
[0079]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[0080]
(First embodiment)
A first embodiment of the present invention will be described with reference to FIGS. 1 and 2. FIG. 1 is a flowchart showing a circuit parameter extraction method according to the first embodiment of the present invention, and FIG. 2 is a schematic diagram showing a cross-section of a finished wiring shape according to the size of a space between wirings.
[0081]
First, correlation data 101 between the layout data 100 and the distance between the model wiring and the wiring existing in the same layer and adjacent to the finished width of the model wiring is prepared. Although the correlation data 101 has been described above, as shown in FIG. 2, when the wiring is laid out in a narrow space, even if the wiring is finished in a substantially square shape, the parasitic capacitance between the same-layer wirings is reduced. In addition, when the layout is performed with the wiring interval widened, the width of the bottom part of the wiring is increased due to the characteristics of the wiring etching process. As data, data as shown in FIG. 15 is prepared by formulating it or making a table. A specific example of calculation using numerical values will be described in the fourth embodiment.
[0082]
The correlation data 101 is obtained by actual measurement data or process simulation. The presence / absence of optical proximity effect correction and its specifications are the same as those used when actually creating mask data from layout data. It must be.
[0083]
As an advantage of preparing a table or function expressed as a parameter with the distance between the left and right wirings as a parameter for each wiring width, in the tabulated expression, the desired value prepared based on the actual measurement value etc. at that table point The simulation can be performed by itself, and the approximate calculation expression (normally complemented with a linear or higher order function between the points) enables high-precision fitting at successive points related to the wiring space or the like.
[0084]
Also, as a way of holding data, it is possible to reduce the amount of data rather than preparing data for each different wiring width by expressing the increase / decrease amount of the top and bottom portions at the end of the layout wiring instead of individually for the wiring width. good.
[0085]
Further, when the wiring width to be used is limited to some extent as in the ASIC design method, the same layer is used instead of the wiring finishing width depending on the layout distance between the same layer wirings 101 and the wiring finishing width. Dependency data between the wiring layout distance and the finished wiring resistance value may be used.
[0086]
In the case where the layout with a fixed wiring width is mainly used as in the ASIC design method, the circuit simulation with less calculation amount can be performed by directly using the resistance value of the wiring of interest.
[0087]
In FIG. 1, first, the wiring width, the wiring length, and the distance between the wirings are extracted from the layout data 100 by the extraction process 102 as in the conventional method. Since the inter-wiring distance here is also used for capacity calculation, it includes the extraction of the distance to the different layer metal.
[0088]
Next, with respect to the wiring width extracted in the step 102, the same distance between the wirings in the same layer extracted in the step 102 and the inter-wiring distance dependency data 101 of the wiring finished width prepared in advance are used. The wiring finish width is predicted by correcting the layout width of the wiring to be performed (step 103). By using the corrected expected finished wiring width and distance between the wirings and the wiring capacity data 104, it is possible to calculate the wiring resistance and the wiring capacity as close to the actual finished values after the process step with high accuracy ( Step 105).
[0089]
According to the present embodiment, by using the correlation data between the finished width of the analysis wiring that has not been considered in the prior art and the distance between the analysis wiring and the wiring existing around the analysis wiring, the wiring resistance / Capacitance extraction and highly accurate circuit simulation are possible. As a result, it is possible to suppress timing errors such as timing skew that may be caused by a difference in wiring resistance value caused by a difference in wiring distance at each location on the chip.
[0090]
In addition, since this method can be used including the change in the wiring finish width caused by the lithography process, even if the wiring top portion dimension is not finished as the mask layout pattern, the difference is also possible. In addition, wiring parameters can be extracted and simulated with high accuracy. As an application method, the resist used for wiring is OK level that does not lead to disconnection or short-circuit according to the processing rule, but the standard value for the finish variation is required to have an accuracy that cannot be realized without using OPC. In such a case, by using the circuit parameter extraction considering the inter-wiring distance and the wiring finish width dependency of the present invention without OPC, a high-accuracy high-priced mask that can handle fine additional patterns for OPC and a long-time OPC mask No processing is required.
[0091]
(Second Embodiment)
A second embodiment of the present invention will be described with reference to FIGS. FIG. 3 is a flowchart showing a circuit parameter extraction method according to the second embodiment of the present invention, and FIG. 4 is a flowchart in which a part of FIG. 3 is changed. Note that the present embodiment relates to a method of utilizing wiring area ratio dependency data of the wiring finish width.
[0092]
Further, not only the wiring area ratio but also the case where the inter-wiring distance dependency data 101 of the wiring finish width shown in the first embodiment is used together will be described as an example.
[0093]
First, the flowchart of FIG. 3 will be described.
[0094]
In FIG. 3, a step 102 for extracting a wiring width, a wiring length, and a distance between wirings from the layout data 100, a wiring interval-wiring finished width correlation data 101 to be referred to, and a finished wiring width correction prediction process 103 using the correlation data 101. Is the same as in the first embodiment.
[0095]
In the present embodiment, correlation data 201 between the wiring area ratio and the wiring finish width as shown in FIG. 13 is newly prepared. Based on the layout data 100 and the wiring data extracted therefrom, the wiring area ratio is calculated by the wiring area ratio calculating step 200. In the wiring area ratio calculation step 200, ideally, it is desirable to obtain the area ratio while appropriately dividing the target process step into a region range affected by the peripheral pattern. In the etching process, it is desirable to calculate the area ratio within a range of about 10 times the minimum processing size. When dealing with a reduction in the wiring film thickness such as dishing and erosion in the metal CMP process used in the damascene process, Although it depends on the hardness of the polishing pad and the like, it is desirable to handle within the range of 100 to 500 μm.
[0096]
However, since calculating the area ratio within a certain limited range for all the wirings of interest has a large CAD processing load, as long as there is no significant density in the chip, it is as described in claim 11. In addition, the area ratio may be calculated and used for the target block unit or the entire chip.
[0097]
With reference to the correlation data 201 between the wiring area ratio and the wiring finished width, the area ratio obtained in the wiring area ratio calculating step 200 is referred to the corrected predicted wiring width value obtained in the finished wiring width correction predicting step 103. Further, the wiring width is corrected (step 202). By using the wiring width and the wiring interval calculated through the above steps, the wiring capacitance data 104 is referred to, and the final finished predicted values of the wiring resistance and the wiring capacitance are obtained in Step 105.
[0098]
The method shown in FIG. 3 is expressed by a flowchart in which the correction process is divided into two steps of wiring interval dependency and wiring area ratio dependency. However, as shown in step 203 of FIG. If the same-layer wiring interval-wiring finish width correlation data using the wiring area ratio as a parameter is prepared, the flow can be simplified. (However, as a concept, FIG. 3 and FIG. 4 are the same.)
The advantages of the area ratio calculation range to be handled are summarized as follows.
[0099]
(1) When the area ratio is calculated by taking a region affected by the process step of interest when calculating the area ratio, if the area ratio is calculated in the range of about 10 times the minimum processing dimension in the etching process, the process finish Therefore, it is possible to extract circuit parameters with high accuracy.
[0100]
(2) When the area ratio is calculated, if the area ratio with respect to the chip area is used, the area ratio calculation can be performed only once. Therefore, the wiring area ratio is taken into consideration without performing high-load CAD processing. Circuit parameter extraction is possible.
[0101]
(3) When the area ratio is calculated, the area ratio with respect to the block area is used, and the area ratio can be calculated several times to several tens of times. Therefore, wiring can be performed without performing high-load CAD processing. Circuit parameters can be extracted in consideration of the area ratio.
[0102]
At the block level, (1) there are many or few wirings, (2) whether a large number of transistors with a large gate width are used, or (3) which of the transistors and wirings determines the area. This is an effective method when the CAD load is also taken into account because it is often determined.
[0103]
As described above, it is possible to extract wiring resistance and wiring capacitance parameters with high accuracy by using the correlation data between the wiring layer area ratio of the analysis wiring and the finished width of the analysis wiring. In addition, it is possible to suppress timing errors such as timing skew that may be caused by differences in wiring resistance values caused by wiring density differences or wiring distance differences at various locations on the chip.
[0104]
(Third embodiment)
A third embodiment of the present invention will be described with reference to FIGS.
[0105]
In the present embodiment, a method of handling the wiring cross-sectional shape as a trapezoid will be mainly described. Furthermore, both the dependency of the wiring finish width on the same-layer wiring distance and the area ratio shown in the first and second embodiments are handled, and correction for the resist finish width such as optical proximity correction is added. Show.
[0106]
FIG. 5 is a flowchart showing a circuit parameter extraction method according to the third embodiment of the present invention. In FIG. 5, reference numeral 301 denotes inter-wiring distance-wiring finish width correlation data using an area ratio as a parameter, measured from an actual semiconductor diffused using a mask subjected to correction such as optical proximity correction for the resist. is there.
[0107]
Even if there is no allowance in the lithography capability of the mask and the resist width itself before etching varies depending on the distance between wirings, an almost desired value can be realized as the resist finished dimension by applying OPC correction to the mask data. In general, when the distance between the patterns approaches a certain distance, the resist finished pattern becomes thin due to light interference. Therefore, in the OPC correspondence, the mask of such a proximity pattern is corrected to be large. When a certain distance is left, the resist dimensions are finished according to the mask pattern dimensions. However, this time, if the pattern becomes an isolated pattern, the resist tends to become thin due to the resist over-etching. However, the resist etching is often performed by a chemical solution. If the resist pattern is patterned to a desired pattern size in the exposure process, the resist etching of the isolated pattern is generally not a problem.
[0108]
Even if the resist is finished to a desired value by utilizing the OPC as described above, the present invention has a problem that only the top portion of the wiring cross section is finished and the distance between the wirings is increased. According to this, the bottom part becomes thicker. FIG. 6 schematically shows the cross-sectional shape of the wiring formed by the dry etching process.
[0109]
FIG. 7 is created based on the actual measurement values based on the schematic diagram of FIG. 6. The graph (a) shows the finished width with respect to the distance between the wirings, and the difference between the top and the bottom on one side with respect to the distance between the wirings. It is a graph (b) which shows. As shown in FIG. 7A, when the distance between the wirings is about 6 times the minimum space, the bottom becomes almost twice the size of the top. Assuming that the wiring cross-sectional shape is trapezoidal and the wiring film thickness is 0.4 μm, the wiring resistance ratio per unit length is estimated to be about 1.5 times that of a square with a width according to the resist dimensions. It becomes the resistance value of.
[0110]
Incidentally, increasing the distance between the adjacent same-layer wirings is a design technique that is often used to reduce the wiring capacity and increase the speed. FIG. 8B shows the inter-wiring distance and the difference in the finished value between the top and bottom around one side based on the data in FIG.
[0111]
Returning to FIG. 5 again, the design flow will be described in order.
[0112]
First, logic and timing design is performed (step 300), and the wiring width, the wiring length, and the inter-wiring distance including the same layer / different layers are extracted from the layout data 100 created based on the net list (step 102). For the extracted wiring width, as shown in FIG. 8B, the inter-wiring distance dependency is used for each of the right end / left end of the wiring, and the finished width per unit length is calculated. The wiring resistance and the wiring capacitance are obtained according to the cross-sectional area (step 105). This wiring resistance calculation is still highly accurate when divided into a plurality of portions having different distances from adjacent wirings in the length direction.
[0113]
Thereafter, using the wiring resistance and capacitance calculated with high accuracy in step 105, the final timing verification simulation is performed in step 304 to confirm the operation. In the timing verification result determination step 306, if the timing verification is NG The circuit for timing adjustment and the layout correction are performed in step 303 so as to be OK again, and the process returns to the resistance value / capacitance value calculation step 105. When the timing is OK in step 306, the resist size correction step 302 of the same processing as that used when collecting the data 301 is performed, and then the mask data is processed in step 303 to generate mask data. This is the design procedure.
[0114]
As described above, by treating the wiring cross section as a trapezoidal shape, it is possible to perform resistance calculation with higher accuracy than those handled as a conventional square cross section. The top side of the trapezoidal shape is determined by the lithography accuracy of the resist for wiring, but according to this embodiment, it is possible to accurately handle the taper shape generated by wiring dry etching and the thickness of the bottom in a form closer to the real thing. It becomes.
[0115]
In addition, by expressing the correlation data with the distance between wiring and the wiring area ratio only by the thickness of the lower base relative to the upper base, the amount of preparation data can be reduced rather than preparing correlation data for each wiring thickness. I can do it.
[0116]
Furthermore, by using together the measures such as the optical proximity effect correction for correcting the change in the finished width of the resist caused by lithography and the resist thinning for the isolated pattern, a high-precision simulation can be realized with a value close to the design value.
[0117]
Next, the method of handling the wiring cross-sectional shape according to the present embodiment as a trapezoid will be described using more specific numerical examples.
[0118]
As shown in FIG. 8 (a), the wiring width is 0.4 μm, the wiring length is 1 mm, and the analysis wiring has a distance of 0.9 μm and a distance of 1.5 μm. It is assumed that they are arranged. The wiring cross-sectional shape is treated as a trapezoidal shape. In addition, when the wiring film thickness is 0.4 μm and the wiring interval on the layout is 0.5 μm and exists on both sides, the wiring top portion width is 0.4 μm and the wiring bottom portion width is 0.5 μm.
[0119]
Here, the resistance value R [Ω] of the wiring represents the wiring cross-sectional area of A [μm. ² ] If the wiring length is L [mm],
(Equation 1)
R [Ω] = (0.18 / A) × L [mm] × 200 [Ω / mm]
It is assumed that it is given in inverse proportion to the wiring cross-sectional area A and in proportion to the wiring length L.
[0120]
When the distance from the adjacent wiring is S [μm], the amount of change in the top portion and the bottom portion is constant with no increase / decrease as shown in FIG. As the distance increases, the width becomes thicker up to a distance of 1.5 μm, and the width of the thickening does not increase beyond the distance of 1.5 μm. Although FIG. 8B is a table format notation, it may be expressed by a function expression as shown below.
[0121]
Now, since the distance between the analysis wiring and the left adjacent wiring is 0.9 μm, the bottom width increment is 0.1 μm from FIG. 8B. Further, since the distance between the analysis wiring and the right adjacent wiring is 1.5 μm, the bottom width increment is 0.25 μm from FIG. 8B. Therefore, the top width of the analysis wiring is 0.4 μm, and the bottom width is 0.5 + 0.1 + 0.25 = 0.85 μm.
[0122]
Since the wiring cross-sectional area can be calculated by the formula for calculating the trapezoidal area (upper base + lower base) × height ÷ 2, (0.4 + 0.85) × 0.4 ÷ 2 = 0.25 [μm ² ].
[0123]
Therefore, by using the above formula 1, the resistance value of the wiring is calculated as R = (0.18 / 0.25) × 1 × 200 = 144 [Ω].
[0124]
When the change due to the distance between wiring is not taken into account, the cross-sectional area is always 0.18 μm. ² Since the wiring resistance is 200 [Ω], there is a resistance value error of {(200-144) / 200} × 100 = 28%.
[0125]
In this way, the wiring cross section is expressed in a trapezoidal shape, the correlation data between the distance between the analysis wiring and the adjacent wiring and the analysis wiring finished width, and the table notation or function notation can be used to improve the wiring resistance parameter extraction accuracy. This can be further improved.
[0126]
(Fourth embodiment)
A fourth embodiment of the present invention will be described with reference to FIG.
[0127]
FIG. 9 is a flowchart showing a circuit parameter extraction method according to the fourth embodiment of the present invention. The process up to step 105 is based on the third embodiment. In addition, the shape assumed when the resistance value calculated in step 105 was used in the logic and timing design 300 before the actual layout (in this case, approximately the square shape realized in the normal line and space). Step 701 for detecting the difference in comparison with the wiring resistance value in the case of a close cross-sectional shape), and when there is a deviation more than a specified value, the wiring width with respect to the original layout data 100 A determination process 702 for determining that the correction process should be performed, and a process 703 for actually performing a correction process on the layout wiring width.
[0128]
By processing based on this flow, the analysis target wiring again approaches the resistance value assumed at the time of the logic and timing design of the process 300, so the verification result in the final timing verification simulation process 304 is more It becomes easy to become OK. In addition, it is possible to suppress timing errors such as timing skew that may be caused by differences in wiring resistance values caused by differences in wiring density and wiring distances at various locations on the chip.
Incidentally, the latest automatic placement and routing tool has a function of laying out with the same length of wiring from the clock distribution point to each flip-flop in order to align the wiring resistance and suppress the clock skew. Even if such a function is provided, this case is an effective skew suppression method because it is not taken into consideration when a difference in wiring resistance occurs due to the density of the wiring or the distance between the wirings.
[0129]
(Fifth embodiment)
A fifth embodiment of the present invention will be described with reference to FIG.
[0130]
FIG. 10 is a schematic diagram showing an assumed process variation range of the gate length.
[0131]
As shown in FIG. 10, the main design margin assumed at the time of LSI design has an important factor of process variation in addition to power supply voltage variation and temperature dependency. Of the process fluctuations, the most effective for the operation speed of the LSI is usually the drive current (Ids) of the transistor, and the dominant factor of the variation amount of the Ids is the gate length. In the case of the ASIC design method, a desired function is realized by combining parts (libraries) prepared at the cell level. As a result, there are various layouts corresponding to the number of corresponding products, and all of the many layouts. On the other hand, since the normal operation of the LSI must be guaranteed, it is inevitably assumed that the processing accuracy of the gate electrode is widened to some extent. However, since the gate electrode area ratio is determined at the stage when the layout is actually finished, the assumed variation range can be set narrow after the layout is determined.
[0132]
Since the pattern of the gate electrode is very fine, the influence of the dimensional accuracy of the pattern forming resist is significant. For this reason, the effect of the resist finish size is greater than the etching of the electrode material itself, and when the gate electrode area ratio is small, the resist becomes overetched and the gate length is finished narrower. In this case, the threshold voltage of the transistor exhibiting the short channel effect in the region where the gate length is thin is lowered, the transistor drive current value is increased, and the operating speed is high. As a result, the guaranteed value in terms of speed can be set to a high value.
[0133]
On the other hand, when the gate electrode area ratio is large, the resist becomes under-etched and the gate length is increased. In this case, the threshold voltage is increased in a transistor having a short channel effect in a region where the gate length is thin, and the speed margin is reduced although it is within the range of the gate width management standard. If any process failure such as an increase in delay occurs, the risk of yield loss due to lack of speed increases.
[0134]
As an advantage when the gate length is large, there is a margin with respect to the standby current standard which is an important performance item in portable devices and the like because the off-state current of the transistor is reduced. In particular, the minimum gate length is often used in the layout of an LSI. When the standby current is laid out using a minimum gate length transistor, the best and worst conditions for the average condition are about It is generally 1/10 times and about 10 times the variation. For this reason, by taking this area ratio into consideration, it becomes possible to estimate the standard value by eliminating the area ratio dependency of the gate length variation, and the guaranteed standard value of the standby current value considering the process variation can be increased. It becomes possible to set to a small value. In other words, when the gate length is fine, the risk for an increase in off-current is estimated in advance, so that the Vt implantation condition is reset to a higher level under process diffusion conditions even after the layout is completed. Can be dealt with at an early stage.
[0135]
(Sixth embodiment)
A sixth embodiment of the present invention will be described with reference to FIG.
[0136]
In this embodiment, the concept of the finished width of the gate electrode material and its area ratio dependency described in the fifth embodiment is used for each block.
[0137]
Since current microprocesses have been miniaturized and the degree of integration has increased, it is common to have a plurality of functional blocks in one chip. For this block, (1) when the block area is determined by wiring, (2) when the block area is determined by the bulk part before the wiring process, (3) when the gate electrode area is very high like a memory cell, etc. There is a tendency that the area ratio of the gate electrode is different for each block.
[0138]
FIG. 11 is a schematic diagram showing an LSI block arrangement according to the sixth embodiment of the present invention.
[0139]
In FIG. 11, it is assumed that the block A (902) has a small gate area ratio, the block B (903) has a large gate area ratio, and the block C (904) has a gate area ratio between the blocks A and B. As the original design policy, the speed variation variation from the average due to process variation is assumed to be + 30% on the high speed condition side and + 30% on the slow condition side, and the assumed speed ranges of blocks A, B, and C are respectively It is assumed that block A is + 30% to -20%, block B is + 20% to -30%, and block C is + 25% to -25%. Since the block A has only to see -20% instead of -30% on the slow side, there is a margin in speed. For a general-purpose microprocessor or the like whose chip value changes according to the frequency, the added value increases as the speed increases. For example, the block A (block A (904) in FIG. Between the signal 905 coming from 902) and the signal 906 coming from the block B (903), which arises from the density of the gate electrodes for each block, which cannot be detected as long as the transistor finish dimensions in the LSI are the same. The transistor drive current difference and the timing skew resulting therefrom can be detected.
[0140]
Conventionally, all transistors formed on an LSI chip are treated as being formed under the same conditions, and timing verification is performed by simulation or the like. However, in reality, there are differences in the finished width of the gate length due to the difference in the gate electrode area ratio, and the resulting differences in transistor drive capability and circuit operation speed. By executing the timing verification simulation in consideration, it is possible to design an LSI with an optimum margin setting for each block.
[0141]
Further, when the target LSI (900) has an asynchronous design, it can occur between the signal 905 from the block A (902) and the signal 906 from the block B (903) for the block C (904). Timing skew can be handled with high accuracy, and the design quality of the LSI can be improved.
[0142]
(Seventh embodiment)
A seventh embodiment of the present invention will be described with reference to FIG.
[0143]
In the present embodiment, the concept of the finished width of the gate electrode material and its area ratio dependency described in the sixth embodiment is further applied to the standard cell level.
[0144]
FIG. 12 is a schematic diagram showing the standard cell library layout of LSI in the seventh embodiment of the present invention.
[0145]
In FIG. 12, standard cell units 907 are connected in the horizontal direction to form a standard cell lane 908, and a circuit block is formed in such a manner that a plurality of standard cell lanes of the same structure are arranged in the vertical direction, although they are different from the used cells. ing. Actually, the horizontal width of each standard cell differs in circuit configuration depending on its function and driving capability, and naturally, the area and horizontal width of each cell are also different.
[0146]
In order to more accurately handle the difference in the finished width of the gate length due to the difference in the gate electrode area ratio, the gate electrode area ratio is calculated after the block is divided into the region range affected by the gate electrode formation process.
[0147]
Ideally, it is more accurate to calculate the gate electrode area ratio for all the cells by scanning the cells 1A, 1B, 1C. However, in order to reduce the load of the CAD calculation, the standard cell block is divided into a size of the region range affected by the gate electrode formation process, and the gate electrode area ratio is calculated for each divided block to calculate the calculated area ratio. Can be used for all cells present in the region.
[0148]
As a result, the difference in the finished width of the gate length due to the difference in the gate electrode area ratio, and the difference in the transistor driving capability and the difference in the circuit operation speed caused by the difference in the gate electrode area ratio are handled in the area range of the entire block with respect to the process formation process. It is possible to handle the process phenomenon more faithfully and in detail, and the simulation accuracy is improved.
[0149]
Therefore, it is possible to design with an appropriate gate width and cell size, and it is possible to suppress timing skew due to a difference in driving current between cells which may occur due to different gate electrode area ratios. LSI design is possible.
[0150]
(Eighth embodiment)
An eighth embodiment of the present invention will be described with reference to FIGS.
[0151]
Also in STI (Shallow Trench Isolation), since an etching process is used when forming an isolation portion of STI, a taper due to etching is generated at the boundary between the isolation region and the active region.
[0152]
FIG. 17A is a plan view and FIG. 17B is a cross-sectional view illustrating an STI structure when the distance between transistor active regions is short. FIG. 18 is a plan view (a) and a cross-sectional view (b) for illustrating the STI structure when the distance between the transistor active regions is large.
[0153]
When the distance between the transistor active regions as shown in FIG. 18 is increased from the state shown in FIG. 17, the taper angle attached to the STI increases. However, although it depends on the etching conditions, at the time of STI formation, since the resist exists on the transistor active region, the top side that is easy to finish according to the resist dimensions of the wiring pattern determines the transistor gate width. The shape is not significantly affected by the Taber shape at the separation end.
[0154]
However, if the channel is formed slightly below the surface, or if the STI etching conditions are set such that the top side under the resist is narrower, the layout data is used for simulation. When extracting circuit data, the difference between the mask layout width and the finished width of the transistor gate width caused by the arrangement distance between the transistors cannot be ignored.
[0155]
Therefore, it is possible to extract transistor parameters with higher accuracy by correcting and extracting the difference between the mask layout width and the finished width of the transistor gate width caused by the distance between the transistors when extracting the gate width. Therefore, it is possible to improve the accuracy of the timing verification simulation.
[0156]
Next, an example using specific numerical values will be described with reference to FIGS.
[0157]
FIG. 19 is a plan view (a) and a cross-sectional view (b) schematically showing a state where two transistors are adjacent to each other with an STI separation on the left and right of the analysis transistor. FIG. 20 shows the correlation between the inter-transistor distance and the gate width penetration from the end on the mask layout.
[0158]
As shown in FIG. 19, the wiring width on the analysis transistor layout is 0.4 μm, and the adjacent transistor is spaced by a distance of 0.7 μm on the left side and a distance of 1.5 μm on the right side in the direction in which the gate width extends. Shall exist.
[0159]
The transistors are separated by STI, and the ends thereof are tapered according to the distance between adjacent transistors. Whether or not the gate width of the transistor to be analyzed is affected depends on the etching conditions, but in this example, the conditions are such that the intermediate point between the top and bottom of STI isolation is etched according to the layout width. Assuming that when the distance between adjacent transistors is 0.5 μm, a finished edge of the electrical gate width of the transistor is formed at the layout boundary portion of the transistor active region, and the distance between the adjacent transistors increases as shown in FIG. It is assumed that such a transistor gate width penetration amount ΔWg. Although FIG. 20 is represented in a table format, it may be represented by a function expression.
[0160]
When the layout shown in FIG. 19 is performed, the distance between the analysis transistor and the left adjacent transistor is 0.7 μm. Therefore, as shown in FIG. 20, the amount of penetration from the left side of the transistor gate width is 0.01 μm, Since the distance between the analysis transistor and the right adjacent transistor is 1.5 μm, the penetration amount from the right side of the transistor gate width is 0.05 μm.
[0161]
Therefore, the finished gate width of the analysis transistor is 0.4−0.01−0.05 = 0.34 μm.
[0162]
When the change due to the distance between adjacent transistors existing in the direction in which the gate width extends is not considered, the transistor gate width is 0.34 / 0, assuming that the transistor gate width is always 0.4 μm according to the layout dimension. .4 = 0.85, which includes an error of 15%.
[0163]
Thus, by considering the change in the gate width of the analysis transistor due to the distance between adjacent transistors in the direction in which the gate width of the transistor to be analyzed extends, the extraction accuracy of the transistor gate width can be further improved. become.
[0164]
【The invention's effect】
As described above, according to the present invention, the following effects are listed.
[0165]
(1) By using the correlation data between the finished width of the target wiring and the distance between the wiring around the target wiring, high-precision wiring resistance / capacitance extraction and high-precision circuit simulation are possible. As a result, it is possible to suppress timing errors such as timing skew that may be caused by a difference in wiring resistance value caused by a difference in wiring distance at each location on the chip.
[0166]
(2) It can be used including the change in the wiring finish width caused by the lithography process. As an application method, the resist used for the wiring is OK level that does not lead to disconnection or short-circuit according to the processing rule, but the standard value regarding the finish variation is required to have an accuracy that cannot be realized without using OPC. Even in this case, by using the accurate circuit parameter extraction considering the inter-wiring distance and the wiring finish width dependency of the present invention without OPC, a high-accuracy high-cost mask that can handle fine additional patterns for OPC and long-time OPC Mask processing becomes unnecessary.
[0167]
(3) By using the correlation data between the wiring layer area ratio of the analysis wiring and the analysis wiring finished width, it is possible to extract wiring resistance and wiring capacitance parameters with high accuracy. In addition, it is possible to suppress timing errors such as timing skew that may be caused by differences in wiring resistance values caused by wiring density differences or wiring distance differences at various locations on the chip.
[0168]
(4) By treating the wiring cross section as a trapezoidal shape, it is possible to perform resistance calculation with higher accuracy than that handled as a conventional square cross section. The top side of the trapezoidal shape is determined by the lithography accuracy of the wiring resist, but it is possible to accurately handle the tapered shape and the bottom thickness generated by the wiring dry etching in a form closer to the real thing. In addition, by expressing the correlation data with the distance between wiring and the wiring area ratio only by the thickness of the lower base relative to the upper base, the amount of preparation data can be reduced rather than preparing correlation data for each wiring thickness. I can do it. Furthermore, by using together the measures such as the optical proximity effect correction for correcting the change in the finished width of the resist caused by lithography and the resist thinning for the isolated pattern, a high-precision simulation can be realized with a value close to the design value.
[0169]
(5) It is possible to estimate the standard value by eliminating the area ratio dependency of the gate length variation, and it is possible to set the guaranteed standard value of the standby current value considering the process variation to a smaller value. . In other words, when the gate length is fine, the risk for an increase in off-current is estimated in advance, so that the Vt implantation condition is reset to a higher level under process diffusion conditions even after the layout is completed. Can be dealt with at an early stage.
[0170]
(6) By performing the timing verification simulation in consideration of the density of the gate area for each block, it is possible to design an LSI with an optimum margin setting for each block.
[0171]
(7) Design with an appropriate gate width and cell size is possible, and timing skew due to a difference in driving current between cells that can occur due to different gate electrode area ratios can be suppressed, resulting in high quality. LSI design is possible.
[0172]
(8) It is possible to further improve the extraction accuracy of the transistor gate width by considering the change in the gate width of the analysis transistor due to the distance between adjacent transistors in the direction in which the gate width of the analysis target transistor extends. Become.
[Brief description of the drawings]
FIG. 1 is a flowchart showing a circuit parameter extraction method according to a first embodiment of the present invention.
FIG. 2 is a schematic diagram showing a cross-section of the finished shape of the wiring according to the size of the space between the wirings.
FIG. 3 is a flowchart showing a circuit parameter extraction method according to a second embodiment of the present invention.
FIG. 4 is a flowchart in which a part of FIG. 3 is changed.
FIG. 5 is a flowchart showing a circuit parameter extraction method according to a third embodiment of the present invention.
FIG. 6 is a schematic diagram showing a cross-sectional shape of a wiring formed by a dry etching process.
FIG. 7 is a graph (a) showing the finished width with respect to the distance between the wirings, and a graph (b) showing the difference between the top and the bottom on one side with respect to the distance between the wirings.
FIG. 8B is a wiring layout and cross-sectional view according to the third embodiment of the present invention, and FIG. 8B is a diagram showing correlation data of the inter-wiring distance and the finished value difference between the top and bottom around one side.
FIG. 9 is a flowchart showing a circuit parameter extraction method according to the fourth embodiment of the present invention.
FIG. 10 is a schematic diagram showing an assumed process variation range of the gate length according to the fifth embodiment of the present invention.
FIG. 11 is a schematic diagram showing a block arrangement of an LSI according to a sixth embodiment of the present invention.
FIG. 12 is a schematic view showing an arrangement of standard cell libraries of an LSI according to a seventh embodiment of the present invention.
FIG. 13 is a cross-sectional view showing a wiring manufacturing process
FIG. 14 is a cross-sectional view (a) of a line and space pattern, a cross-sectional view (b) of an isolated pattern, and an enlarged cross-sectional view (c).
FIG. 15 is a graph showing the correlation between the wiring distance and the wiring finish width;
FIG. 16 is a graph showing the correlation between the wiring area ratio and the wiring finish width;
17A and 17B are a plan view and a cross-sectional view showing an STI structure when the distance between transistor active regions is short
18A and 18B are a plan view and a cross-sectional view showing an STI structure when the distance between the transistor active regions is long (FIG. 18A).
FIGS. 19A and 19B are a plan view and a cross-sectional view schematically showing a state in which two transistors are adjacent to each other with an STI isolation sandwiched between the right and left sides of an analysis transistor for explaining an eighth embodiment of the present invention. b)
FIG. 20 is a diagram showing correlation data between the distance between transistors and the gate width penetration amount from the end on the mask layout.
[Explanation of symbols]
100 layout data
101 Correlation data between wiring interval and wiring finish width
102 Wiring width, wiring length, wiring distance extraction process
103 Wiring width (and wiring interval) correction process
104 Wiring interval-wiring capacity correlation data
105 Wiring resistance and wiring capacitance calculation process
200 Wiring area ratio calculation process
201 Wiring area ratio-wiring finish width correlation data
202 Wiring width (and wiring interval) correction process
203 Wiring interval-wiring finish width correlation data considering the wiring area ratio
300 Logic & Timing Design Process
301 Wiring interval-wiring finish width correlation data taking into account the wiring area ratio (with correction of optical proximity effect etc. on resist)
302 Resist dimensional control correction process such as optical proximity correction
303 Mask data processing step
304 Final timing verification simulation process
305 Timing adjustment circuit and layout correction process
306 Timing verification result determination step
701 Wiring resistance value comparison before and after correction, and difference detection process
702 Necessity judgment process of layout wiring width correction
703 Layout wiring width correction process
900 Target LSI
901 I / O cell part
902 Block A
903 Block B
904 Block C
905 Signal from block A to block C
906 Signal from block B to block C
907 Standard Cell Library Unit
908 Standard Cell Lane
909 Area ratio calculation range of gate electrode pattern

Claims (18)

A method for extracting circuit parameters such as wiring resistance and wiring capacitance from a layout of a semiconductor integrated circuit,
Preparing correlation data between the distance between the model wiring and the wiring of the same layer existing around the model wiring and the difference between the mask layout width and the finished width of the model wiring;
From the actual layout, extract the wiring length and wiring width of the analysis wiring, and extract the distance between the analysis wiring and the wiring existing in the same layer,
The wiring finish width obtained by referring to the correlation data with respect to the extracted layout wiring width of the analysis wiring and the distance between the extracted analysis wiring and the wiring existing around the analysis wiring. A circuit parameter extraction method characterized by calculating a wiring resistance value and a wiring capacitance value. A method for extracting circuit parameters such as wiring resistance and wiring capacitance from a layout of a semiconductor integrated circuit,
Prepare the correlation data between the wiring pattern area ratio, which is the ratio of the wiring pattern existing in the set area of the wiring layer of the model wiring, and the amount of change in the finished width of the model wiring,
Extract the wiring length and wiring width of the analysis wiring from the actual layout, calculate the pattern area ratio of the wiring layer of the analysis wiring,
Changes in the wiring finish width due to the area ratio obtained by referring to the correlation data with respect to the extracted wiring length and wiring width of the analysis wiring and the calculated pattern area ratio of the analysis wiring layer A circuit parameter extraction method characterized by calculating a wiring finish value by using a quantity and calculating a wiring resistance value and a wiring capacitance value. Said correlation data to prepare expresses the wire cross section of the model wiring, rectangle, square with acute and obtuse corner of the non-rectangular, square or polygonal, as the circumferential curve of the curvature in the shape of a side Express as a change in shape and any shape that represents a side with a function curve,
In addition to the extracted wiring length and wiring width on the layout of the analysis wiring, the wiring resistance value per unit length is calculated by calculating the cross-sectional area of the finished wiring cross-sectional shape obtained by referring to the correlation data The circuit parameter extracting method according to claim 1, wherein the resistance value and the capacitance value of the analysis wiring are calculated from the wiring resistance value per unit length and the wiring length on the layout.   The circuit parameter extracting method according to claim 3, wherein the wiring cross-sectional shape is handled as a trapezoidal shape. A method for extracting circuit parameters such as wiring resistance and wiring capacitance from a layout of a semiconductor integrated circuit,
Correlation expressing the distance between the model wiring and the same-layer wiring existing around the model wiring, and the amount of change in the resistance value and the capacitance value of the model wiring, using the wiring width on the layout of the model wiring as a parameter Prepare the data
From the actual layout, extract the wiring length and wiring width of the analysis wiring, and extract the distance between the analysis wiring and the wiring existing in the same layer,
A circuit parameter extraction method characterized in that a finished resistance value and a capacitance value of the analysis wiring are calculated by referring to the correlation data using the extracted layout wiring width of the analysis wiring as a parameter. A method for extracting circuit parameters such as wiring resistance and wiring capacitance from a layout of a semiconductor integrated circuit,
With the wiring width on the layout of the model wiring as a parameter, the wiring pattern area ratio that is the ratio of the wiring pattern existing in the setting region where the wiring layer of the model wiring is present, and the amount of change in the resistance value and capacitance value of the model wiring Prepare correlation data for
Extract the wiring length and wiring width of the analysis wiring from the actual layout, calculate the pattern area ratio of the wiring layer of the analysis wiring,
A circuit parameter extraction method characterized in that a finished resistance value and a capacitance value of the analysis wiring are calculated by referring to the correlation data using the extracted layout wiring width of the analysis wiring as a parameter. It said correlation data to prepare a circuit parameter extracting method of any one of claims 1, which is a table or an approximate equation of 6.   The circuit parameter extraction method according to claim 2 or 6, wherein when calculating the wiring pattern area ratio, the area ratio of the wiring pattern occupying the entire chip area is calculated.   7. The circuit parameter extraction method according to claim 2, wherein the wiring pattern area ratio is calculated for each circuit block when the wiring pattern area ratio is calculated.   7. The circuit according to claim 2, wherein when calculating the wiring pattern area ratio, the pattern area ratio is calculated within a range in which a wiring processing process step can be influenced by an area ratio of a wiring pattern existing around. Parameter extraction method.   The circuit parameter extraction method according to claim 10, wherein the wiring pattern area ratio is calculated within a region range 10 times the minimum processing dimension.   The circuit parameter extraction method according to any one of claims 1 to 11, and an optical proximity effect correction for correcting a change amount of a resist finish width caused by a lithography process in a photomask data generation process used in a semiconductor lithography process, and A method for designing a semiconductor integrated circuit, which is used in combination with correction of the amount of change in resist thinning in an isolated pattern caused by a resist etching process. 12. The wiring resistance value is calculated by referring to the mask layout data and calculating the wiring finish width, wiring resistance value, and wiring capacitance value by using the circuit parameter extraction method according to claim 1. Compare the wiring capacitance value with the desired wiring resistance value and wiring capacitance value set at the time of design,
It is determined whether or not to correct the layout wiring width to approximate the desired wiring resistance value and wiring capacitance value,
As a result of the determination, when correction is necessary, the layout wiring width is corrected so as to approach the desired wiring resistance value and wiring capacitance value,
A method of designing a semiconductor integrated circuit, wherein timing simulation verification of the semiconductor integrated circuit is performed again using the wiring resistance value and wiring capacitance value after correcting the layout wiring width. A method for designing a semiconductor integrated circuit, comprising:
Treat the cross-sectional shape of the wiring as a trapezoidal shape,
Performs optical proximity correction that corrects the amount of change in the resist finish width caused by the lithography process, and layout correction that adds the amount of change in resist thinning in the isolated pattern caused by the resist etching process.
Along with the first correlation data of the distance between the model wiring and the same-layer wiring existing around the model wiring, and the upper and lower bases of the finished shape of the model wiring, using the layout correction, Preparing the second correlation data of the area ratio of the wiring layer of the model wiring and the finished values of the top and bottom of the finished shape of the model wiring using the layout correction;
From the actual layout, extract the wiring length and wiring width of the analysis wiring, and extract the distance between the analysis wiring and the wiring existing in the same layer,
Calculate the pattern area ratio of the wiring layer of the analysis wiring,
Using the wiring length of the extracted analysis wiring, the wiring width, the distance between the analysis wiring and the wiring existing in the same layer, and the pattern area ratio of the wiring layer of the analysis wiring, Calculate the amount of change in the upper and lower bases in the trapezoidal cross section, the wiring cross-sectional area of the analysis wiring, and the wiring resistance value,
It is determined whether or not the calculated analysis wiring resistance value is to be corrected to approach the desired wiring resistance value and wiring capacitance value intended at the time of design,
If correction is necessary as a result of the determination, a correction is made to change the layout wiring width so as to approach the desired wiring resistance value and wiring capacitance value,
A method of designing a semiconductor integrated circuit, wherein a timing verification simulation of the semiconductor integrated circuit is performed again using the wiring resistance value and wiring capacitance value after the correction. 15. The method for designing a semiconductor integrated circuit according to claim 14, wherein the area ratio of the wiring pattern occupying the entire chip area is calculated when calculating the pattern area ratio of the wiring layer of the analysis wiring . In calculating the pattern area ratio of the wiring layer of said analysis lines, a method of designing a semiconductor integrated circuit according to claim 14, wherein calculating the area ratio of the wiring pattern for each circuit block. 15. When calculating the pattern area ratio of the wiring layer of the analysis wiring, the pattern area ratio is calculated within a range in which the wiring processing process can be affected by the area ratio of the wiring pattern existing in the periphery. The semiconductor integrated circuit design method described. 18. The method of designing a semiconductor integrated circuit according to claim 17 , wherein the pattern area ratio of the wiring layer of the analysis wiring is calculated within a region range of 10 times the minimum processing dimension.

Images