Semiconductor integrated circuit having layout in which buffers or protection circuits are arranged in concentrated manner

1. Field of the Invention

The present invention relates to a semiconductor integrated circuit, and more particularly to a layout in a semiconductor chip.

2. Description of the Background Art

Recently, from a viewpoint of cost reduction, reduction in a chip area has been demanded in a semiconductor integrated circuit such as a microcomputer. In addition, higher operation speed and lower power consumption have also been demanded in a semiconductor integrated circuit.

For example, Japanese Patent Laying-Open No. 08-125130 discloses a semiconductor integrated circuit aiming to stabilize a circuit operation by reducing crosstalk or noise due to capacitive coupling between signal lines located in different interconnection layers in a semiconductor integrated circuit having a multilayer interconnection. Having a plurality of metal interconnection layers, the semiconductor integrated circuit includes a signal line as well as a ground line and a power supply line arranged in parallel to each other on respective opposing sides of the signal line in an identical interconnection layer and fixed to a ground potential and a power supply voltage potential respectively. The semiconductor integrated circuit is characterized in that the signal line, the ground line and the power supply line are arranged proximate to each other to such an extent that a spacing therebetween is set to at most a thickness of an interlayer insulating layer electrically insulating interconnections located in different interconnection layers.

In the conventional semiconductor integrated circuit, a pad and an I/O buffer (or a protection circuit such as an off-transistor and a diode) are arranged proximate to each other in a periphery of a chip. When the pad and the I/O buffer (or the protection circuit) are arranged in such a manner, reduction in a chip size is restricted.

For example, if a semiconductor integrated circuit implements a microcomputer, circuits such as a CPU (Central Processing Unit), an ROM (Read Only Memory), an RAM (Random Access Memory), and the like are mounted on a chip. Here, each circuit is constituted of elements such as an MOS (Metal Oxide Semiconductor) transistor or a capacitor. As these elements are reduced in size through microfabrication, scale of each circuit can be made smaller.

On the other hand, the size of the pad or the I/O buffer (or the protection circuit) cannot be made smaller, along with size reduction of the circuit such as a CPU. As the I/O buffer or the protection circuit protects an internal circuit (such as a CPU, an ROM, an RAM, and the like) against noise or surge introduced from the outside, sufficient area thereof is required. In the conventional semiconductor integrated circuit, the I/O buffer or the protection circuit is arranged in proximity to the pad. Therefore; if a protection circuit having a large area is provided in a region between pads or a region between the pad and the CPU, length of four sides of the chip is determined by the pad and the protection circuit, which prevents reduction in chip size.

Meanwhile, if the area of the pad or a spacing between the pads is significantly made smaller, defects in manufacturing may increase in an assembly step such as dicing or wire bonding. Therefore, an area of the pad or a pitch between the pads cannot significantly be changed in an attempt to reduce the chip size.

In preparing a broader product range in order to meet various demands from customers, generally, product variety is reinforced by varying a capacity of the RAM or the ROM, without substantially modifying the CPU. Even if the CPU, the RAM and the ROM are arranged so as not to leave an empty space in the semiconductor chip in a certain product, however, there is an empty space created in the chip in a new product provided with an RAM having a size smaller than in the conventional product. In such a product as well, the I/O buffer or the protection circuit has been arranged proximate to the pad.

In summary, such measures as-reduction in size of a circuit element for reducing the chip size or arrangement of the CPU, the RAM and the ROM so as not to leave an empty space have conventionally been taken. Once the chip size is determined, however, it has not been easy to reduce the chip size in the case of change in the size of the RAM or the ROM.

An object of the present invention is to provide a semiconductor integrated circuit capable of achieving reduction in a chip size.

In summary, a semiconductor integrated circuit according to the present invention includes: a central processing unit; a non-volatile memory; a volatile memory; a plurality of buffers or a plurality of protection circuits; a plurality of pads; and a plurality of metal interconnections. The non-volatile memory stores information on processing performed in the central processing unit in a non-volatile manner. The volatile memory temporarily stores the information. The plurality of buffers or the plurality of protection circuits are arranged in a region in a main region including the central processing unit, the non-volatile memory and the volatile memory, except for the central processing unit, the non-volatile memory and the volatile memory, on a main surface of a semiconductor substrate. The plurality of pads are provided corresponding to the plurality of buffers or the plurality of protection circuits, respectively. The plurality of metal interconnections directly connect the plurality of buffers or the plurality of the protection circuits to corresponding ones of the plurality of pads, respectively.

Therefore, a primary advantage of the present invention is that the I/O buffers or the protection circuits are arranged in a concentrated manner in the region on the surface of the semiconductor substrate, except for the region for the CPU, the RAM, the ROM, and the like, so that a distance between the pads or a spacing between the pad and the circuit such as the CPU can be made smaller, thereby reducing the chip size.

The foregoing and other objects, features, aspects and advantages of the present invention will become more apparent from the following detailed description of the present invention when taken in conjunction with the accompanying drawings.

FIG. 1 shows a layout of a semiconductor integrated circuit in a first embodiment.

FIG. 2 shows a layout of a semiconductor integrated circuit according to an arrangement method in related art different from the first embodiment.

FIG. 3 shows a layout of the semiconductor integrated circuit according to a variation of the first embodiment.

FIG. 4 shows a layout of a semiconductor integrated circuit in a second embodiment.

FIG. 5 shows a layout of a semiconductor integrated circuit in a third embodiment.

FIG. 6 is a cross-sectional view along the line VI-VI in FIG. 5 .

FIG. 7 shows a layout of a semiconductor integrated circuit in a fourth embodiment.

FIG. 8 is a cross-sectional view along the line VIII-VIII in FIG. 7 .

FIG. 9 schematically shows a method of manufacturing a sidewall SW in FIG. 8 .

FIG. 10 shows a layout of a semiconductor integrated circuit in a fifth embodiment.

An embodiment of the present invention will be described hereinafter in detail with reference to the drawings. The same or corresponding elements have the same reference characters allotted.

First Embodiment

FIG. 1 shows a layout of a semiconductor integrated circuit in a first embodiment. Referring to FIG. 1 , a semiconductor integrated circuit 1 includes a central processing unit (denoted as CPU in the drawing) 2 , a non-volatile memory 4 storing information on prescribed processing performed in central processing unit 2 in a non-volatile manner, and a volatile memory 6 temporarily storing the information. Non-volatile memory 4 implements, for example, a flash memory. -Volatile memory 6 implements, for example, an SRAM (Static Random Access Memory).

Semiconductor integrated circuit 1 further includes buffers B 1 to B 9 serving as I/O buffers. Buffers B 1 to B 9 are arranged in a concentrated manner in a single region SP 1 in a main region MS on a surface of a semiconductor substrate, except for the region for central processing unit 2 , non-volatile memory 4 and volatile memory 6 .

Semiconductor integrated circuit 1 further includes pads P 1 to P 9 provided corresponding to I/O buffers B 1 to B 9 respectively and metal interconnections L 1 to L 9 directly connecting buffers B 1 to B 9 to pads P 1 to P 9 respectively.

Semiconductor integrated circuit 1 has a multilayer interconnection. Metal interconnections L 1 to L 9 are formed in an identical interconnection layer. Each of metal interconnections L 1 to L 9 is provided in an interconnection layer higher than any of an interconnection layer where a metal interconnection included in central processing unit 2 is provided, an interconnection layer where a metal interconnection included in non-volatile memory 4 is provided, and an interconnection layer where a metal interconnection included in volatile memory 6 is provided, relative to the surface of the semiconductor substrate.

Specifically, metal interconnections L 3 to L 5 are provided above a plurality of bit lines BL 1 included in non-volatile memory 4 . Metal interconnections L 6 , L 7 are provided above a plurality of bit lines BL 2 included in volatile memory 6 . Metal interconnections L 8 , L 9 are provided in the interconnection layer above signal lines SL 1 , SL 2 included in central processing unit 2 .

Characteristics of a configuration of semiconductor integrated circuit 1 according to the first embodiment will be summarized. Buffers B 1 to B 9 are arranged in a concentrated manner in region SP 1 distant from pads P 1 to P 9 . Region SP 1 is a region in main region MS except for central processing unit 2 , non-volatile memory 4 and volatile memory 6 . As the buffer requiring a large area is not provided around the pad, a spacing between the pads or a spacing between the pad and the internal circuit (such as central processing unit 2 ) can be made smaller, thereby reducing the chip size.

In the semiconductor integrated circuit according to the present invention, a length of the metal interconnection connecting the pad to the buffer is longer than in the conventional semiconductor integrated circuit. Therefore, the semiconductor integrated circuit according to the present invention may be more susceptible to noise or surge due to larger interconnect resistance between the pad and the buffer. Such a problem, however, can be solved by forming each of metal interconnections L 1 to L 9 in a top interconnection layer.

In general, in a multilayer interconnect structure, a metal interconnection in an upper layer has higher degree of freedom-in design with regard to a width or a thickness of an interconnection. Therefore, if each of metal interconnections L 1 to L 9 is arranged in the top interconnection layer, the width or thickness thereof can be set to attain sufficiently small interconnect resistance between the pad and the buffer.

FIG. 2 shows a layout of a semiconductor integrated circuit according to a different arrangement method in related art. Referring to FIG. 2 , a semiconductor integrated circuit 11 is different from semiconductor integrated circuit 1 in FIG. 1 in including buffers BA to BD provided proximate to the pad. As the configuration of semiconductor integrated circuit 11 is otherwise the same as that of corresponding portions of semiconductor integrated circuit 1 , description thereof will not be repeated.

A region SP 4 is provided inside semiconductor integrated circuit 11 . In a certain semiconductor integrated circuit, even if region SP 4 is not provided, it is created by reducing the size of non-volatile memory 4 or by reducing the size of volatile memory 6 .

In the different layout, even if an empty space having an area sufficient for concentrated arrangement of buffers is provided on the surface of the semiconductor substrate, reduction in the chip size has been restricted, because the buffers or the protection circuits have been provided proximate to the pad.

In addition, in the layout where the pads are arranged along four sides of the chip, it is not possible to arrange the buffer or the protection circuit around pads PA to PD provided in four corners respectively. Namely, the empty space created around the pad in the corner has also been a factor to restrict reduction in the chip size.

Meanwhile, in semiconductor integrated circuit 1 in FIG. 1 , buffers B 1 to B 9 are arranged in region SP 1 in a concentrated manner. Accordingly, by efficiently utilizing the empty space in the chip, the chip is made smaller. It is noted that the-buffer and the protection circuit may be provided in region SP 1 in FIG. 1 . Alternatively, the protection circuit may be provided in a region between the pads or in a region between the pad and the internal circuit. Specifically, the protection circuit may be provided in a region SP 2 between pad P 1 and pad P 2 or a region SP 3 between pads P 5 , P 6 and non-volatile memory 4 , volatile memory 6 .

FIG. 3 shows a layout of the semiconductor integrated circuit according to a variation of the first embodiment. Referring to FIG. 3 , a semiconductor integrated circuit 1 A is different from semiconductor integrated circuit 1 in FIG. 1 in including protection circuits PR 1 to PR 9 provided in region SP 1 , instead of buffers B 1 to B 9 . The protection circuit is represented, for example, by an off-transistor or a diode. As the configuration of semiconductor integrated circuit 1 A is otherwise the same as that of corresponding portions of semiconductor integrated circuit 1 , description thereof will not be repeated. As can be seen in the variation of the first embodiment, the chip size can be reduced also by arranging the protection circuit in the empty space in the chip.

In the variation of the first embodiment, buffers B 1 to B 9 may be provided in region SP 1 or in region SP 2 or SP 3 .

As described above, according to the first embodiment, the buffers or the protection circuits are arranged in a concentrated manner in the empty space on the surface of the semiconductor substrate, so that the chip size of the semiconductor integrated circuit can be reduced.

Second Embodiment

FIG. 4 shows a layout of a semiconductor integrated circuit in a second embodiment. Referring to FIG. 4 , a semiconductor integrated circuit 21 is different from semiconductor integrated circuit 1 in FIG. 1 in including a plurality of regions SP 5 , SP 6 instead of region SP 1 . In the second embodiment, the buffers (or the protection circuits) are arranged in regions SP 5 , SP 6 as blocks. FIG. 4 shows blocks BC 1 , BC 2 arranged in regions SP 5 , SP 6 respectively.

As the configuration of semiconductor integrated circuit 21 is otherwise the same as that of corresponding portions of semiconductor integrated circuit 1 , description thereof will not be repeated.

In the first embodiment, the buffers or the protection circuits are arranged in a concentrated manner in a single empty space. Therefore, if the area of the single empty space is smaller than the sum of the areas of the buffers or the protection circuits, it is impossible to arrange the buffers or the protection circuits in the empty space. In such a case, the chip size cannot be reduced, because the buffers or the protection circuits are arranged around the pads according to the different arrangement method.

In the semiconductor integrated circuit according to the second embodiment, so long as the total area of the plurality of empty spaces is greater than the sum of the area of the buffers or the protection circuits in spite of a small area of each of the plurality of empty spaces, the buffers or the protection circuits are arranged in each of the plurality of empty spaces so that the empty space inside the chip can efficiently be made use of. Therefore, the chip size can be reduced as in the first embodiment.

As described above, according to the second embodiment, the buffers or the protection circuits are arranged in a concentrated manner in the plurality of empty spaces inside the chip, so that the chip size can be reduced.

Third Embodiment

FIG. 5 shows a layout of a semiconductor integrated circuit in a third embodiment. Referring to FIG. 5 , a semiconductor integrated circuit 31 is different from semiconductor integrated circuit 1 in FIG. 1 in further including dummy interconnections DL 1 to DL 8 serving as metal interconnections formed in the interconnection layer where metal interconnections L 1 to L 9 are provided, and supplied with a prescribed potential. As the configuration of semiconductor integrated circuit 31 is otherwise the same as that of corresponding portions of semiconductor integrated circuit 1 , description thereof will not be repeated. It is noted that the buffers or the protection circuits in semiconductor integrated circuit 31 are shown as a block BC 3 provided in region SP 1 .

The metal interconnection among the plurality of metal interconnections connecting the pad to the buffer (or the protection circuit) includes a portion extending in parallel to any one of the plurality of bit lines BL 1 or any one of the plurality of bit lines BL 2 , and dummy interconnections DL 1 to DL 8 are provided so as to sandwich, from sides, the parallel portion of the metal interconnection. In FIG. 5 , dummy interconnections DL 1 , DL 2 are provided so as to sandwich from sides a portion of metal interconnection L 3 , that extends in parallel to bit line BL 1 . Similarly, dummy interconnections DL 3 , DL 4 are provided with regard to a portion of metal interconnection L 4 that extends in parallel to bit line BL 1 , and dummy interconnections DL 5 , DL 6 are provided with regard to a portion of metal interconnection L 5 that extends in parallel to bit line BL 1 . In addition, dummy interconnections DL 7 , DL 8 are provided with regard to a portion of metal interconnection L 6 that extends in parallel to bit line BL 2 .

In the semiconductor integrated circuit according to the present invention, a part of the metal interconnection connecting the pad to the buffer or a part of the metal interconnection connecting the pad to the protection circuit is arranged above the non-volatile memory or the volatile memory. In such a case, as a result of variation in a potential of the bit line of the non-volatile memory or of the bit line of the volatile memory provided in a lower layer in response to a signal transmitted through the metal interconnection provided in the upper layer, a phenomenon (crosstalk) in which erroneous information is read from the non-volatile memory or the volatile memory takes place.

Here, crosstalk is caused in the following manner. Namely, as the metal interconnection in the upper layer is arranged in parallel to the bit line in the lower layer, an interlayer capacitance between these interconnection layers locally becomes large. In the third embodiment, the interlayer capacitance is reduced by arranging dummy interconnections supplied with the prescribed potential on opposing sides of the metal interconnection in the upper layer, so that occurrence of crosstalk can be suppressed.

The prescribed potential supplied to the dummy interconnection can readily be set, if the prescribed potential is the power supply potential or the ground potential. Here, the ground potential is particularly preferred. In general, a ground region formed of metal and supplied with the ground potential is provided around the pad. Therefore, the dummy interconnection can readily be arranged by connecting the dummy interconnection to the ground region.

If a P-type substrate is used as the semiconductor substrate, capacitive component other than line-to-line capacitance is present between the metal interconnection and the dummy interconnection, as a result of connection of the dummy interconnection to the substrate for setting a substrate potential to the ground. Noise caused in the signal transmitted through the metal interconnection tends to be absorbed by the capacitive component other than the, line-to-line capacitance. Therefore, in order to suppress occurrence of crosstalk, the potential of the dummy interconnection is preferably set to the ground.

FIG. 6 is a cross-sectional view along the line VI-VI in FIG. 5 . FIG. 6 shows bit lines BL 1 A to BL 1 C of non-volatile memory 4 as well as metal interconnection L 3 and dummy interconnections DL 1 , DL 2 . Metal interconnection L 3 and dummy interconnections DL 1 , DL 2 are provided in the interconnection layer above bit lines BL 1 A to BL 1 C, with an insulating film LA 1 being interposed. In addition, an insulating film LA 2 is provided in order to cover metal interconnection L 3 and dummy interconnections DL 1 , DL 2 . If metal interconnection L 3 (and dummy interconnections DL 1 , DL 2 ) serves as the uppermost interconnection, insulating film LA 2 serves as a protection film covering the surface of the chip.

There is interlayer capacitance present between metal interconnection L 3 and bit line BL 1 A. The interlayer capacitance is implemented by a surface-to-surface capacitance Cs between an upper surface of bit line BL 1 A and a lower surface of metal interconnection L 3 and a fringe capacitance Cf between a side surface of bit line BL 1 A and a side surface of metal interconnection L 3 .

In addition, there is line-to-line capacitance Cc present between metal interconnection L 3 and each of dummy interconnections DL 1 , DL 2 . Here, line-to-line capacitance Cc is a capacitance between the side surface of metal interconnection L 3 and a side surface of the dummy interconnection.

In general, relation of Q=CV holds, where Q represents charges, C represents capacitance, and V represents a potential. Therefore, if the interlayer capacitance is great, variation in the potential of metal interconnection L 3 induces variation in the potential on bit line BL 1 A.

In a general memory, information read from the memory cell is produced as a potential difference between two bit lines connected to that memory cell. The potential difference between the two bit lines is sensed and amplified by a sense amplifier, to attain a logic level of either binary 1 or 0. Therefore, if the potential of one of the two bit lines is varied due to variation in the potential of metal interconnection L 3 as a result of great interlayer capacitance, the potential difference produced between the two bit lines is sensed and amplified by the sense amplifier, which results in reading of erroneous information from the memory. In addition, as the capacitance is greater, a time constant becomes greater, which also leads to delay in the signal on the bit line.

Line-to-line capacitance Cc and fringe capacitance Cf implement the * capacitance with regard to the side surface of metal interconnection L 3 . Fringe capacitance Cf can be lowered by increasing line-to-line capacitance-Cc. If fringe capacitance Cf is lowered, the interlayer capacitance is lowered. Therefore, occurrence of crosstalk can be suppressed.

As described above, according to the third embodiment, the dummy interconnections fixed to a prescribed potential are arranged on both sides of the portion of the metal interconnection that extends in parallel to the bit line of the memory, the metal interconnection connecting the pad to the buffer or the protection circuit. Therefore, erroneous information reading from the memory can be prevented and operation reliability can be improved.

Fourth Embodiment

FIG. 7 shows a layout of a semiconductor integrated circuit in a fourth embodiment. Referring to FIG. 7 , as the layout of a semiconductor integrated circuit 41 is similar to that of semiconductor integrated circuit 31 in FIG. 5 , description thereof will not be repeated. The fourth embodiment is different from the third embodiment in that a sidewall formed of a dielectric is provided on each side surface of metal interconnections L 1 to L 9 and dummy interconnections DL 1 to DL 8 .

In the fourth embodiment, the sidewall having a dielectric constant higher than that of the insulating film covering the metal interconnection and the dummy interconnection is provided in the metal interconnection and the dummy interconnection, so that the line-to-line capacitance between the metal interconnection and the dummy interconnection can be higher than in the third embodiment. Meanwhile, the fringe capacitance out of the interlayer capacitance between the metal interconnection and the bit line is lower than in the third embodiment. Since the interlayer capacitance is lower than in the third embodiment, crosstalk is less likely than in the semiconductor integrated circuit according to the third embodiment.

FIG. 8 is a cross-sectional view along the line VIII-VIII in FIG. 7 . Referring to FIG. 8 , a sidewall SW is provided on opposing side surfaces of metal interconnection L 3 and dummy interconnections DL 1 , DL 2 . Sidewall SW is formed of a dielectric. Sidewall SW has a dielectric constant higher than that of insulating film LA 2 . Sidewall SW is composed, for example, of tantalum oxide (Ta 2 O 5 ), hafnium oxide (HfO 2 ) or the like. Meanwhile, insulating film LA 2 is implemented, for example, by a plasma oxide film and more specifically by plasma TEOS (Tetraethylorthosilicate). If insulating film LA 2 serves as a protection film, a plasma nitride film is generally used as the protection film. In other words, a material for forming sidewall SW should be a material having a dielectric constant higher than that of a plasma oxide film or a plasma nitride film.

As the dielectric constant is higher, the capacitance is greater. Comparing the fourth embodiment with the third embodiment, sidewalls SW are provided on both side surfaces of metal interconnection L 3 and dummy interconnections DL 1 , DL 2 , respectively. Therefore, line-to-line capacitance Cc is greater than in the third embodiment. As the line-to-line capacitance is greater, fringe capacitance Cf can be smaller. Therefore, the interlayer capacitance can be smaller than in the third embodiment.

FIG. 9 schematically shows a method of manufacturing sidewall SW in FIG. 8 . Referring to FIG. 9 , initially at step SI, a metal film is deposited on insulating film L 1 A, and a photoresist PH is applied to the metal film. Thereafter, an interconnection pattern is transferred to photoresist PH in a lithography step. In an etching step, the metal film is removed except for a portion covered with photoresist PH, to form metal interconnection L 3 and dummy interconnections DL 1 , DL 2 .

Thereafter at step S 2 , an insulating film LA 3 is formed on metal interconnection L 3 and dummy interconnections DL 1 , DL 2 by plasma CVD (Chemical-Vapor Deposition). This insulating film ultimately serves as sidewall SW. Here, insulating film LA 3 is an insulating film attaining isotropic step coverage.

Thereafter at step S 3 , anisotropic etching mainly including vertical components is carried out, to form sidewall SW.

Thereafter at step S 4 , insulating film LA 2 (or the protection film) covering metal interconnection L 3 , dummy interconnections DL 1 , DL 2 and sidewall SW is formed by plasma CVD.

As described above, according to the fourth embodiment, the sidewall formed of the dielectric is provided on the side surface of the metal interconnection and on the side surface of the dummy interconnection provided above the bit line, so that the interlayer capacitance between the interconnection in the upper layer and the bit line in the lower layer can be lowered. Erroneous information reading from the memory can thus be prevented and operation reliability is improved.

Fifth Embodiment

FIG. 10 shows a layout of a semiconductor integrated circuit in a fifth embodiment. Referring to FIG. 10 , a semiconductor integrated circuit 51 is different from semiconductor integrated circuit 1 in FIG. 1 in including metal interconnections L 3 B, L 4 B, L 5 B, and L 6 B having a zigzag shaped portion instead of metal interconnections L 3 , L 4 , L 5 , and L 6 . As the configuration of semiconductor integrated circuit 51 is otherwise the same as that of corresponding portions of semiconductor integrated circuit 1 , description thereof will not be repeated.

As described already, the interlayer capacitance implemented by the surface-to-surface capacitance and the fringe capacitance is present between the interconnection in the upper layer and the bit line in the lower layer. In the third and fourth embodiments, the interlayer capacitance is made smaller by lowering the fringe capacitance. On the other hand, in the fifth embodiment, the interlayer capacitance is made smaller by lowering the surface-to-surface capacitance.

As an area of an overlapping portion between the bit line in the lower layer and the interconnection in the upper layer is smaller, the surface-to-surface capacitance is lower. Therefore, the interconnection in the upper layer is formed in a zigzag shape so as to minimize an area of a portion in parallel to the bit line in the lower layer. More specifically, the zigzag-shaped portion is formed to bend a plurality of times, in such a manner as changing a bending direction alternately in a first direction in parallel to any one of the plurality of bit lines BL 1 or any one of the plurality of bit lines BL 2 and in a second direction different from the first direction, in the interconnection layer where metal interconnections L 1 to L 9 are formed.

For example, metal interconnection L 4 B will be described as a representative. Metal interconnection L 4 B includes bent portions A 1 , A 2 . Bent portions A 1 , A 2 are formed to bend a plurality of times in such a manner as changing a bending direction alternately in a first direction (L 41 B) in parallel to bit line BL 1 and in a second direction (L 42 B) different from the first direction (a direction perpendicular to bit line BL 1 ). As a result of bent portions A 1 , A 2 , an area of a portion overlapping between metal interconnection L 4 B and one specific bit line is made smaller. As the surface-to-surface capacitance between metal interconnection L 4 B and the bit line in the lower layer is thus made smaller, the interlayer capacitance becomes lower.

For the sake of convenience for description, FIG. 10 shows interconnections L 3 B, L 4 B passing between bit lines BL 1 . In general, however, the metal interconnection connected to the pad is thicker than a width of the bit line and an interval between the bit lines. Therefore, it is impossible to arrange the metal interconnection in the upper layer in a manner not overlapping with any of the two parallel bit lines. Here, according to the fifth embodiment, the metal interconnection in the upper layer has a zigzag shape, so that the portion overlapping with the bit line in parallel can be made smaller.

An angle of bending in the zigzag-shaped portion is not limited to 90°, and any angle except for 0° and 180° may be set.

As described above, according to the fifth embodiment, the interconnection overlapping with the bit line can be in a zigzag-shape so as to make smaller an area of the portion overlapping with the bit line, whereby operation reliability can be improved.

Although the present invention has been described and illustrated in detail, it is clearly understood that the same is by way of illustration and example only and is not to be taken by way of limitation, the spirit and scope of the present invention being limited only by the terms of the appended claims.