High-performance low-power 3D SRAM unit and process integration process

By using fin or nanosheet transistors in SRAM cells, adjusting the number of fins or nanosheet width ratios, and combining CFET technology, the challenge of stacked SRAM cells is solved in different levels, achieving high performance and low power compatibility Sex and flexibility.

CN118900554APending Publication Date: 2024-11-05INTERUNIVERSITAIR MICRO ELECTRONICS CENT (IMEC VZW) +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410538435.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2023-05-04
Filing Date
2024-04-30
Publication Date
2024-11-05

AI Technical Summary

Technical Problem

The existing stacked SRAM cells have insufficient optimization of layout design and interconnect solutions during the manufacturing process, especially the difficulty in meeting the needs of high-performance (HP) and low-power (LP) SRAM at the same time, and different manufacturing levels are made. There are challenges in number or width of transistors.

Method used

Using a fin or nanosheet-based transistor design, adapting to HP and LP SRAM needs by adjusting the fin number or nanosheet width ratio, and combining complementary field effect transistor (CFET) technology to support the manufacturing of sequential and monolithic CFETs. Transistor design at different levels is achieved using processes such as trench cutting and isotropic trimming.

Benefits of technology

A small footprint SRAM cell design is realized, which improves the speed of HP SRAM and the read stability of LP SRAM, provides flexible process flow for different types of SRAM, supports process flow outside the A3 node, and meets different performances. need.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118900554A_ABST
    Figure CN118900554A_ABST
Patent Text Reader

Abstract

The present disclosure relates to a three-dimensional (3D) static random access memory (SRAM), and provides a 3DSRAM cell and a method for processing a transistor of the 3D SRAM. The SRAM cell includes two PU transistors arranged in a first level, two PD transistors arranged in a second level located above or below the first level, and two PG transistors arranged in the first level or the second level. The PU transistor and the PD transistor form a pair of cross-coupled inverters. In a first alternative, each transistor is a fin transistor, each PU transistor has a first number of fins, each PD transistor has a second number of fins, and if the PG transistor is arranged in a first level, the ratio of the first number to the second number is 2: 1, or if the PG transistor is arranged in a second level, the ratio of the second number to the first number is 2: 1. The ratio of the first number to the second number is 1: 2.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to three-dimensional (3D) static random access memory (SRAM). The present disclosure provides a 3D SRAM cell and a method for processing transistors of the 3D SRAM cell. The present disclosure particularly provides solutions for both high performance (HP) SRAM and low power (LP) SRAM. Background Art

[0002] SRAM is a semiconductor memory widely used in electronics, microprocessors, and general computing applications. SRAM can store data in a static manner and does not need to be dynamically updated like other types of memory. SRAM includes multiple SRAM cells, each of which is configured to store one bit of data. A typical six-transistor (6T) SRAM cell has four transistors for storing the bit, namely two pull-up (PU) and two pull-down (PD) transistors, which are configured as two cross-coupled inverters. The cross-coupled inverters have two stable states, which determine the logical "0" and "1" states of the bit. In addition to the four transistors for storing the bit, a typical 6T SRAM cell also includes two additional transistors (called transfer gate (PG) transistors), which are used to control access to these four transistors during a bit read or bit write operation.

[0003] Currently, many methods of manufacturing SRAM are aimed at reducing the cell area of ​​SRAM cells and improving their performance. For example, a stacked SRAM cell, in which the transistors of the SRAM cell are arranged in multiple tiers (or levels) stacked one on top of another, can result in a reduced cell area. For example, a stacked SRAM cell can be manufactured using complementary field effect transistor (CFET) technology, in which n-type metal oxide semiconductor (NMOS) transistors and p-type metal oxide semiconductor (PMOS) transistors are processed together in a stacked manner (monolithically or sequentially), and can then be cross-coupled to form two cross-coupled inverters.

[0004] However, the layout design and interconnection solutions for such stacked SRAM cells, as well as the process flow for manufacturing transistors of such SRAM cells, still need to be optimized, especially in order to achieve satisfactory solutions for different kinds of SRAMs (eg, HP SRAM and LP SRAM). Summary of the invention

[0005] Therefore, an object of the present disclosure is to provide a layout design and interconnect solution for stacked 6TSRAM cells for HP SRAM and LP SRAM. Another object is to provide a method for processing transistors of SRAM cells, wherein the method will be integrated with the process flow for manufacturing SRAM.

[0006] Yet another object is to implement SRAM cells with nanosheet-based transistors or fin-based transistors. Therefore, a specific object is to address the challenge of manufacturing different fin-based transistors or nanosheet-based transistors (e.g., transistors with different numbers of fins or different nanosheet widths) in different levels of the SRAM cell. Another object is to implement the SRAM cell using CFET technology. Therefore, the method should support both sequential CFET and monolithic CFET. For sequential CFET, processing may be simpler because the top and bottom levels are manufactured separately and then combined. However, the top and bottom levels are combined together in a monolithic CFET approach. Yet another object is to develop The process flow outside the technology node (referred to as "A3 node").

[0007] These and other objects are achieved by the solutions disclosed herein as described in the independent claims. Advantageous embodiments are further described in the dependent claims.

[0008] A first aspect of the present disclosure provides a 3D SRAM cell, comprising: two PU transistors arranged in a first level of the SRAM cell;

[0009] two PD transistors arranged in a second level of the SRAM cell, the second level being arranged above or below the first level;

[0010] wherein the two PU transistors and the two PD transistors form a pair of cross-coupled inverters; and

[0011] two PG transistors arranged in the first level or the second level;

[0012] wherein in the SRAM cell: each transistor is a fin transistor, each PU transistor has a first number of fins, each PD transistor has a second number of fins, and if the PG transistors are arranged in the first level, the ratio of the first number to the second number is 2:1, or if the PG transistors are arranged in the second level, the ratio of the first number to the second number is 1:2; or

[0013] Each transistor is a nanosheet-based transistor, each PU transistor has a first nanosheet width, each PD transistor has a second nanosheet width, and if the PG transistor is arranged in the first level, the ratio of the first nanosheet width to the second nanosheet width is 2:1, or if the PG transistor is arranged in the second level, the ratio of the first nanosheet width to the second nanosheet width is 1:2.

[0014] As an example, the SRAM cell of the first aspect may be a monolithic 3D SRAM cell. The fin number ratio or nanosheet width ratio is suitable for designing SRAM cells for HP SRAM or LP SRAM, wherein the fin number or nanosheet width of the PG transistor may be different for HP SRAM and LP SRAM, as will be described below.

[0015] A first aspect of the present disclosure provides a layout design and interconnect solution, in particular a stacked 6T SRAM cell for HP SRAM or LP SRAM. The SRAM cell according to the first aspect can be implemented using a nanosheet-based transistor or a fin transistor. The fin transistor can be a fin field effect transistor (FinFET). The nanosheet-based transistor can be a nanosheet transistor or a fork-sheet transistor.

[0016] It is noteworthy that in the present disclosure, the terms "below" and "above", "bottom" and "top" or similar terms should be interpreted relative to each other. Specifically, these terms describe the opposite sides of the SRAM cell, or the opposite sides of any element of the SRAM cell. These terms can describe the relationship of the elements (e.g., transistors) of the SRAM cell along the stacking direction. The stacking direction can be aligned with the arrangement of the first and second levels of the SRAM cell. That is, two (or more) levels are arranged above each other, which means that these levels are arranged one after another along a certain direction (stacking direction).

[0017] Furthermore, in the present disclosure, the term "transistor" does not necessarily refer only to fully processed and functional transistors, but also to intermediate transistor structures, which may include the channel structure of the final transistor (e.g., nanosheets or fins), but may not (yet) include a gate or source / drain (SD) contacts, for example. That is, the term "transistor" in various aspects, implementations, and detailed embodiments of the present disclosure includes intermediate transistor structures.

[0018] In an implementation of the 3D SRAM cell, each PG transistor has a third number of fins, and the third number is equal to the first number or the second number; or each PG transistor has a third nanosheet width, and the third nanosheet width is equal to the first or second nanosheet width.

[0019] The third number of fins or the third nanosheet width, respectively, may be selected depending on whether the SRAM cell is designed for a HP SRAM or a LPSRAM, as described in the following embodiments.

[0020] In one implementation of a 3D SRAM cell, the SRAM cell is used for HP SRAM, and: if the PG transistor is arranged in a first level, the ratio of the first number to the third number and the second number is 2:2:1, or if the PG transistor is arranged in a second level, the ratio of the first number to the third number and the second number is 1:2:2; or if the PG transistor is arranged in the first level, the ratio of the first nanosheet width to the third nanosheet width and the second nanosheet width is 2:2:1, or if the PG transistor is arranged in the second level, the ratio of the first nanosheet width to the third nanosheet width and the second nanosheet width is 1:2:2.

[0021] In one implementation of a 3D SRAM cell, the SRAM cell is used for LP SRAM, and: if the PG transistor is arranged in a first level, the ratio of the first number to the third number and the second number is 2:1:1, or if the PG transistor is arranged in a second level, the ratio of the first number to the third number and the second number is 1:1:2; or if the PG transistor is arranged in the first level, the ratio of the first nanosheet width to the third nanosheet width and the second nanosheet width is 2:1:1, or if the PG transistor is arranged in the second level, the ratio of the first nanosheet width to the third nanosheet width and the second nanosheet width is 1:1:2.

[0022] SRAM cells for HP SRAM have faster speeds because the read path is stronger—due to stronger PG and PD transistors in the case of a 1:2:2 ratio and NMOS PG transistors, or due to stronger PG and PU transistors in the case of a 2:2:1 ratio and PMOS PG transistors. SRAM cells for LP SRAM have better read stability, which provides higher VDD scaling potential for low-power designs.

[0023] In one implementation of the 3D SRAM cell, each PD transistor in the second level is stacked directly above or below one of the PU transistors in the first level; and / or each pair of transistors consisting of a PU transistor in the first level and a PD transistor in the second level is CFET-based.

[0024] Hence, the design of the SRAM cell of the first aspect is compatible with the use of CFET technology.Both monolithic and sequential CFET approaches can be used.

[0025] In one implementation of the 3D SRAM cell: if the PG transistor is arranged in a first level, the PG transistor and the PU transistor in the first level are PMOS transistors, and the PD transistor in the second level is an NMOS transistor; or if the PG transistor is arranged in a second level, the PG transistor and the PD transistor in the second level are NMOS transistors, and the PU transistor in the first level is a PMOS transistor.

[0026] In one implementation of the 3D SRAM cell, a cross-coupling structure for the pair of cross-coupled inverters is arranged in the second level, directly above or below each PG transistor.

[0027] Thus, the cross-coupled connections of the PU and PD transistors can advantageously be formed in the vacated space. This can allow a small footprint for the SRAM cell. The cross-coupled structure can be embedded in a dielectric material that can be arranged in the space above the PG transistor, for example filling the space.

[0028] A second aspect of the present disclosure provides a method for processing transistors of a 3D static random access SRAM cell, the method comprising: forming two PU transistors in a first level of the SRAM cell; forming two PD transistors in a second level of the SRAM cell, the second level being formed above the first level, or the first level being formed above the second level; forming two PG transistors in the first level or the second level; connecting the two PU transistors and the two PD transistors to form a pair of cross-coupled inverters; and wherein: each transistor is a fin transistor, each PU transistor is formed to have a first number of fins, and each PD transistor is formed to have a first number of fins. There is a second number of fins, and if the PG transistor is arranged in the first level, the ratio of the first number to the second number is 2:1, or if the PG transistor is arranged in the second level, the ratio of the first number to the second number is 1:2; or each transistor is a nanosheet-based transistor, each PU transistor is formed to have a first nanosheet width, each PD transistor is formed to have a second nanosheet width, and if the PG transistor is arranged in the first level, the ratio of the first nanosheet width to the second nanosheet width is 2:1, or if the PG transistor is arranged in the second level, the ratio of the first nanosheet width to the second nanosheet width is 1:2.

[0029] A second aspect of the present disclosure provides a process flow suitable for processing transistors of the SRAM cell of the first aspect. The process flow specifically enables the manufacture of different types of fin transistors with different numbers of fins, or different types of nanosheet-based transistors with different nanosheet widths. The process flow can be used in the A3 node.

[0030] In one implementation of the method: each PG transistor is formed to have a third number of fins, and the third number is equal to the first number or the second number; or each PG transistor is formed to have a third nanosheet width, and the third nanosheet width is equal to the first or second nanosheet width.

[0031] In one implementation, the method also includes: forming two intermediate transistors in the second level or the first level, respectively, directly above or directly below the PG transistor in the first level or the second level; removing at least a portion of the channel structure of each intermediate transistor; and forming a cross-coupling structure for a pair of cross-coupled inverters in the space created by removing at least a portion of the channel structure of each intermediate transistor.

[0032] Removing at least a portion of the channel structure of the middle transistor may include removing at least a portion of each fin or nanosheet of the middle transistor. For example, this removal may be achieved by trench cutting, for example, in combination with photolithography. The space created by this removal may advantageously be used to form a cross-coupling structure. This enables the design of an SRAM cell with a small footprint (small cell area).

[0033] For example, removing at least a portion of the channel structure of the middle transistor may include forming an opening (or trench) in a sacrificial gate disposed above the channel structure, and then forming a cut through the channel structure by etching the channel structure from the opening in the sacrificial gate. This may be referred to as trench cutting. The etching may extend completely through the channel structure and may stop before reaching the channel structure of a PG transistor below or above the middle transistor in another level.

[0034] For example, the PG transistor and the middle transistor may be formed by using CFET technology, ie together as a CFET. Furthermore, each pair of PU transistor and PD transistor of the SRAM cell may be formed by using CFET technology, ie together as a CFET as described above.

[0035] In one implementation of the method, all transistors are fin transistors, and the method includes: forming each PD transistor in the second level, whose number of fins is the same as that of the PU transistor in the first level; and the method also includes: if the PG transistor is arranged in the first level, removing at least one fin of each PD transistor so as to reduce the number of fins of the PD transistor to the second number; or if the PG transistor is arranged in the second level, removing at least one fin of each PU transistor so as to reduce the number of fins of the PU transistor to the first number.

[0036] The removal of at least one fin of each PD transistor or each PU transistor can be accomplished by, for example, trench cutting combined with photolithography. According to this implementation, a process flow is provided that allows transistors with different numbers of fins to be manufactured in the top level and the bottom level (first level and second level), in particular when CFET technology is used.

[0037] In one implementation, the method further includes: forming each PG transistor in the first level, whose number of fins is the same as that of the PU transistor, or forming each PG transistor in the second level, whose number of fins is the same as that of the PD transistor; and the method further includes: if the PG transistors are arranged in the first level, removing at least one fin of each PG transistor so as to reduce the number of fins of the PG transistor to the second number; or if the PG transistors are arranged in the second level, removing at least one fin of each PG transistor so as to reduce the number of fins of the PG transistor to the first number.

[0038] The removal of at least one fin of the or each PG transistor may be accomplished by trench cutting. Alternatively, it may be accomplished by active fin cutting prior to any gate patterning. According to this implementation, a process flow is provided that allows transistors with different numbers of fins to be manufactured in the same level.

[0039] In one implementation of the method, all transistors are nanosheet-based transistors, and the method includes: forming each PD transistor in the second level, whose nanosheet width is the same as that of the PU transistor in the first level; and the method also includes: if the PG transistor is arranged in the first level, reducing the nanosheet width of each PD transistor to the second nanosheet width by isotropic trimming; or if the PG transistor is arranged in the second level, reducing the nanosheet width of each PU transistor to the first nanosheet width by isotropic trimming.

[0040] Accordingly, a process flow is provided that allows transistors with different nanosheet widths to be fabricated in the top level and the bottom level (first level and second level), in particular when using CFET technology. Isotropic trimming can be performed with or without photolithography. Isotropic trimming can include isotropically etching the nanosheet (from at least one side) to reduce the nanosheet width.

[0041] In one implementation, the method further includes: forming each PG transistor in the first level, whose nanosheet width is the same as that of the PU transistor in the first level; and the method further includes: if the PG transistor is formed in the first level, reducing the nanosheet width of each PG transistor to a second nanosheet width by isotropic trimming; or if the PG transistor is formed in the second level, reducing the nanosheet width of each PG transistor to the first nanosheet width by isotropic trimming.

[0042] Thus, a process flow is provided that allows transistors with different nanosheet widths to be fabricated in the same level. Trimming can be combined with photolithography.

[0043] In one implementation of the method, the reduction of the number of fins of the PD transistor is performed at or during a replacement metal gate (RMG) process step of a manufacturing process of the SRAM cell, or isotropic trimming is performed accordingly to reduce the nanosheet width of the PD transistor.

[0044] This enables efficient integration of the method of the second aspect into the overall manufacturing process of the SRAM cell.Thus, the method of the second aspect can generally achieve the same advantages as described for the SRAM cell of the first aspect.

[0045] It is worth noting that the present disclosure is not limited to the order in which the respective reductions in the number of fins of, for example, the PG transistor and the PD transistor are performed, or the order in which the various trench cutting steps are performed. Likewise, the present disclosure is not limited to the order in which the various trimmings, for example, the trimming of the nanosheets of the PG transistor and the PD transistor are performed. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The above aspects and implementations are explained in the following detailed description with reference to the accompanying drawings:

[0047] Figure 1 An SRAM cell and an interconnect solution for a HP fin-on-fin CFET SRAM according to the present disclosure are shown.

[0048] Figure 2-3 shows that it can be used for processing Figure 1 FIG. 1 is an exemplary process flow for the PU and PD transistors of an SRAM cell.

[0049] Figure 4 An SRAM cell and an interconnect solution for a LP-fin-on-fin CFET SRAM according to the present disclosure are shown.

[0050] Figure 5An SRAM cell and an interconnect solution for a HP nanosheet-on-nanosheet CFET SRAM according to the present disclosure are shown.

[0051] Figure 6-7 shows that it can be used for processing Figure 5 FIG. 1 is an exemplary process flow for a PG transistor of an SRAM cell.

[0052] Figure 8-9 shows that it can be used for processing Figure 5 FIG. 1 is an exemplary process flow for the PU and PD transistors of an SRAM cell.

[0053] Fig.10 An SRAM cell and an interconnect solution for a HP nanosheet-on-nanosheet CFET SRAM according to the present disclosure are shown.

[0054] Fig.11 An SRAM cell and an interconnect solution for a LP nanosheet-on-nanosheet CFET SRAM according to the present disclosure are shown.

[0055] Figure 12-13 shows that it can be used for processing Fig.10 or Fig.11 FIG. 1 is an exemplary process flow for a PG transistor of an SRAM cell.

[0056] Fig.14 An SRAM cell and an interconnect solution for a HP forksheet-on-forksheet CFET SRAM according to the present disclosure are shown.

[0057] Figure 15-16 shows that it can be used for processing Fig.14 FIG. 1 is an exemplary process flow for the PU and PD transistors of an SRAM cell.

[0058] Fig.17 An SRAM cell and an interconnect solution for a LP fork-on-chip CFET SRAM according to the present disclosure are shown.

[0059] Fig.18 A method for fabricating transistors of any 3D CFET SRAM cell according to the present disclosure is shown. DETAILED DESCRIPTION

[0060] The present disclosure proposes a layout design for a 3D SRAM cell with stacked transistors (i.e., transistors arranged in two different levels). The SRAM cell is particularly a 6T SRAM cell. The SRAM cell includes two PU transistors, two PD transistors, and two PG transistors. The two PU transistors are arranged in the first level of the SRAM cell. Therefore, the two PU transistors can be arranged in the same plane or layer, or, for example, at the same distance from the underlying substrate layer. The two PD transistors are arranged in the second level of the SRAM cell, which is, for example, arranged above or below the first level. For example, if the SRAM cell includes a substrate layer, the first level can be formed directly on the substrate layer, and the second level can be formed directly on the first level and / or above the first level, that is, the distance from the substrate layer is greater than the first level. Alternatively, the second level can be formed directly on the substrate layer, and the first level can be formed directly on the second level and / or above the second level, that is, the distance from the substrate layer is greater than the second level.

[0061] The two PD transistors and the two PU transistors form a pair of cross-coupled inverters, just like in a conventional SRAM cell. The pair of cross-coupled inverters can take two different states (defining "0" or "1") and can therefore store one bit of information, also similar to a conventional SRAM cell. The two PG transistors can be arranged in the first level or in the second level and can allow access to the pair of cross-coupled inverters, which function the same as in a conventional SRAM.

[0062] Typically, if the PG transistors are arranged in the first level, the PG transistors and the PU transistors are PMOS transistors, and the PD transistors are NMOS transistors. However, if the PG transistors are arranged in the second level, the PG transistors and the PD transistors are typically NMOS transistors, and the PU transistors are PMOS transistors.

[0063] Each transistor of the SRAM cell can be a fin transistor (e.g., FinFET) or a nanosheet-based transistor (e.g., a nanosheet transistor or a fork-sheet transistor). In the former case, each transistor has a certain number of one or more fins. In the latter case, each transistor includes one or more nanosheets having a certain width. Generally, according to the concepts of the present disclosure, the PU transistors and the PD transistors have different numbers of fins or different nanosheet widths. Specifically, each PU transistor has a first number of fins and each PD transistor has a second number of fins, or each PU transistor has a first nanosheet width and each PD transistor has a second nanosheet width.

[0064] In the case of using fin transistors, if the PG transistors are arranged in the first level (i.e., in the same level as the PU transistors), the ratio of the first number of fins of the PU transistors to the second number of fins of the PD transistors is 2: 1. Alternatively, if the PG transistors are arranged in the second level (i.e., in the same level as the PD transistors), the ratio of the first number of fins to the second number of fins is 1:2.

[0065] In the case of using nanosheet-based transistors, if the PG transistors are arranged in the first level, the ratio of the first nanosheet width of the PU transistor to the second nanosheet width of the PD transistor is 2: 1. Alternatively, if the PG transistors are arranged in the second level, the ratio of the first nanosheet width to the second nanosheet width is 1:2.

[0066] Each PG transistor may have a third number of fins or a third nanosheet width. The third number of fins of the PG transistor may be equal to the first number of fins of the PU transistor or equal to the second number of fins of the PD transistor. Similarly, the third nanosheet width may be equal to the first nanosheet width of the PU transistor or the second nanosheet width of the PD transistor. The third number of fins or the third nanosheet width depends on whether the SRAM cell is designed for HP SRAM or LP SRAM, respectively.

[0067] In the SRAM cell of the present disclosure, the PD transistor can be formed directly above the PU transistor, or vice versa. For example, each PD transistor in the second level can be stacked directly above or directly below one of the PU transistors in the first level. Thus, each stacked pair including a PU transistor in the first level and a PD transistor in the second level can be manufactured using CFET technology, that is, can be implemented as a CFET. In the present disclosure, CFETs can be processed using sequential or monolithic CFET technology.

[0068] In the following, specific but exemplary embodiments of different SRAM cells according to the present disclosure are described, as well as various interconnect solutions in the SRAM cells. In these embodiments, the transistors of the SRAM cells are based on fin transistors, nanosheet transistors, or fork-sheet transistors, and use CFET technology. Layout designs of different SRAM cells are shown, and the relevant process flows for manufacturing the transistors of the SRAM cells are shown and explained respectively.

[0069] Figure 1 An exemplary SRAM cell 100 for a HP fin-on-fin CFET SRAM according to the present disclosure is shown. That is, Figure 1The SRAM cell 100 is designed for HP SRAM, is based on fin transistors, and can use monolithic CFET technology, for example, for implementing PU and PD transistors. The SRAM cell is shown in a top view, where (a) is a view of the first level 101 and (b) is a view of the second level 102. In this case, the first level 101 is arranged below the second level 102 (closer to the underlying substrate layer).

[0070] because Figure 1 The SRAM cell is for HP SRAM, and since the PG transistors are arranged together with the PU transistors in the first level 101, the ratio of the first number of fins of the PU transistors to the third number of fins of the PG transistors and the second number of fins of the PD transistors in the second level 102 is 2:2:1. That is, in this case, the third number is equal to the first number.

[0071] The fin 1a of the PU transistor and the fin 1a' of the PG transistor form the bottom transistor channels of these transistors, respectively, and can be implemented as silicon channels or other suitable semiconductor material channels. The fin 2a of the PD transistor in the second level 102 forms the top transistor channel, and can also be implemented as a silicon channel or other suitable semiconductor material channel. It is worth noting that in the case of a nanosheet-based SRAM cell 100, these transistor channels will be formed by nanosheets, as shown in the following figures. In short, the fin or nanosheet can form the channel structure of the corresponding transistor.

[0072] Figure 1 The SRAM cell 100 shown in the figure also includes a back metal layer 19 arranged in or below the first level 101, which can be used, for example, to provide a supply voltage (VDD). In addition, the SRAM cell 100 includes a metal one (M1) layer 23 arranged in or above the second level 101, which can be used, for example, to provide a ground voltage (VSS). In addition, the SRAM cell 100 includes a metal intermediate (MINT) layer 22 formed in or above the second level 102, which can be used, for example, to implement word lines (WL), and / or bit lines (BL) and / or bit line bars (BLB).

[0073] Figure 1The SRAM cell 100 may also include a dielectric isolation wall 16, for example, for isolating the SRAM cell 100 from other SRAM cells or the periphery in the SRAM. In addition, similar to a conventional SRAM cell, nodes Q and QB are formed in the SRAM cell 100, and an internal node merge 17 between the top and bottom metal zero (M0) layers may be provided, which may be used to form a cross-coupling between the PU and PD transistors. In addition, the SRAM cell 100 includes a bottom SD epitaxial (Epi) structure 9 (e.g., including p-type silicon germanium) and a top SD epi structure 10 (e.g., including n-type silicon), which may be used to contact the nodes Q and QB.

[0074] In addition, in the SRAM cell 100, at least the via 18 can provide a connection to a supply voltage (VDD) provided by the back metal 19 or the back power rail. In addition, the supervia 20 can provide a connection to the front metal ground voltage (VSS). In addition, the SRAM cell 100 can include a common gate 24, a gate extension 25 for cross-coupling formation, and a spacer merger 21.

[0075] Figure 1 The SRAM cell is designed based on the use of CFET technology, and thus the stacked pair of the PU transistor in the first level 101 and the PD transistor in the second level 102 can be formed as a CFET. Figure 1 The SRAM cell 100 can be manufactured by forming each PD transistor in the second level 102 having the same number of fins 2a as the PU transistor in the first level 101, and then removing at least one fin 2a of each PD transistor to reduce the number of fins 2a of the PD transistor to a second number.

[0076] This can be done by using trench cutting. Specifically, photolithography can be used to process Figure 1 The SRAM cell 100 is provided to implement trench cutting at the RMG process step of the manufacturing process of the SRAM cell. The trench cutting at the RMG process step may mean that the trench cutting is implemented before the functional gate stack is formed (for example, when the dummy gate stack is removed). In this aspect, Figure 2 and 3 shows that it can be used for processing Figure 1 FIG. 1 is an exemplary process flow for the PU and PD transistors of the SRAM cell 100 .

[0077] Figure 2The fin structure in the middle SRAM cell is shown in (a) (as it can be formed after the chemical mechanical polishing (CMP) step of the interlayer dielectric (ILD) 11 in the manufacturing process of the SRAM cell 100). Two different (vertical) cross-sectional views of the middle SRAM cell are shown on the left and right sides. The fin structure is associated with a pair of transistors consisting of a PU transistor in the first (lower) level 101 and a PD transistor in the second (upper) level 102. The fin structure includes a fin 1a of the PU transistor and a fin 2a of the PD transistor. These fins 1a and 2a are separated by an intermediate dielectric isolation (MDI) 3, which can be made of Si3N4, SiCO, SiCON, SiCN or a dual dielectric material such as SiO2 liner plus SiN or SiO2 liner plus SiOCN. The fin structure is exemplarily formed on the substrate layer using a shallow trench isolation (STI) oxide 5. The upper fin 2a of the PD transistor contacts the top SD Epi structure 10, and the lower fin 1a of the PU transistor contacts the bottom SD Epi structure 9.

[0078] The fin structure is surrounded by a dummy gate 6 (e.g., made of amorphous silicon or polycrystalline silicon), wherein the fin structure is separated from the material of the dummy gate 6 by a gate oxide 4. On the dummy gate 6, a gate hard mask 7 is provided, which can be made of Si3N4 and / or SiO2. The dummy gate 6 is also sandwiched by a gate spacer 8, which can be made of Si3N4, SiCO, SiCON, SiBCN or SiOBCN. The fin structure and the dummy gate 6 are further sandwiched by an ILD 11, which can be made of SiO2.

[0079] Figure 2 (b) shows that the dummy gate 6 and the gate hard mask 7 are removed in the next process step.

[0080] Figure 3 In (a) it is shown that in the next process step a spin-on carbon (SoC) coating 12 is applied around the lower fin 1a belonging to the PU transistor and also partially around the MDI 3. The SoC coating 12 may be applied first around the entire fin structure and then may be etched back to expose the upper fin 2a without the need for photolithography.

[0081] Figure 3 In (b) it is shown that in the next process step trench cutting of the top fin 2a of the fin structure belonging to the PD transistor is performed. The trench cutting may include etching into the trench defined by the gate spacer 8 and the ILD 11 and may stop on the MDI 3 acting as an etch stop.

[0082] Figure 3 In a next process step shown in (c), the SoC coating 12 is then stripped (removed).

[0083] Figure 4 An exemplary SRAM cell 400 for a LP fin-on-fin CFET SRAM according to the present disclosure is shown. That is, Figure 4 The SRAM cell 400 is designed for LP SRAM, is based on fin transistors, and can use monolithic CFET technology to implement PU and PD transistors. The SRAM cell 400 is shown in a top view, where (a) is a view of the first level 101 and (b) is a view of the second level 102. The first level 101 is arranged below the second level 102. It is worth noting that Figure 1 and Figure 4 The same elements in FIG. 1 have the same reference numerals, are implemented identically, and are therefore not described again with reference to the above explanation.

[0084] because Figure 4 The SRAM cell 400 is for LP SRAM, and since the PG transistors are arranged together with the PU transistors in the first level 101, the ratio of the first number of fins 1a of the PU transistors to the third number of fins 1a' of the PG transistors and the second number of fins 2a of the PD transistors in the second level 102 is 2:1:1. In other words, the third number is equal to the second number.

[0085] It is worth noting that Figure 4 The SRAM cell can also be used Figure 2 and Figure 3 The process flow shown is such that the number of fins 2a of the PD transistor is reduced to a second number, for example by using trench cutting. The PG transistor may initially be formed with the same number of fins as the PU transistor, and this number may be reduced to the final number of fins 1a' of the PG transistor by trench cutting or by performing active cutting before any (dummy) gate patterning.

[0086] Figure 5 An exemplary SRAM cell 500 for a HP nanosheet-on-nanosheet CFET SRAM according to the present disclosure is shown. That is, Figure 5 The SRAM cell 500 is designed for HP SRAM, based on nanosheet transistors, and can use monolithic CFET technology to implement PU and PD transistors. The SRAM cell 500 is shown in a top view, where (a) is a view of the first level 101 and (b) is a view of the second level 102. The first level 101 is arranged below the second level 102. It is worth noting that Figure 1 and Figure 5 The same elements in FIG. 1 have the same reference numerals, are implemented identically, and are not described again with reference to the above explanation.

[0087] because Figure 5 The SRAM cell 500 is for HP SRAM, and since the PG transistor is arranged together with the PU transistor in the first level 101, the ratio of the first nanosheet width of the nanosheet 1b of the PU transistor to the third nanosheet width of the nanosheet 1b' of the PG transistor and the second nanosheet width of the nanosheet 2b of the PD transistor is 2:2:1. That is, the third nanosheet width is equal to the first nanosheet width.

[0088] Figure 5 The SRAM cell 500 can be manufactured using CFET technology, and thus the PG transistor can be first formed as a CFET. Figure 5 The SRAM cell 500 may be fabricated by forming two middle ("PG") transistors in the second level 102 directly above the PG transistors in the first level 101 and then removing at least a portion of the channel structure of each middle transistor. In addition, a cross-coupled structure for a pair of cross-coupled inverters may be formed in the space created by removing at least a portion of the channel structure of the middle transistors.

[0089] This removal can be done by trench cutting. Specifically, the removal can be done by trench cutting combined with photolithography, and can be achieved at the RMG process step of the manufacturing process of the SRAM cell 500. In this regard, Figure 6 and 7 shows that it can be used for processing Figure 5 1. An exemplary process flow of the PG transistor of the SRAM cell 500 of FIG. 5. It is worth noting that the same or similar process can be used for all other SRAM cells proposed in the present disclosure to manufacture the PG transistor by CFET technology, but ultimately only having the PG transistor in the first level 101.

[0090] Figure 6The nanosheet structure in the middle SRAM cell is shown in (a) (as it can be formed after the CMP step of the ILD 11 in the manufacturing process of the SRAM cell 500). Two different (vertical) cross-sectional views of the middle SRAM cell are shown on the left and right sides. The nanosheet structure is associated with a PG transistor in the first (lower) level 101 and an middle PG transistor in the second (upper) level 102. The nanosheet structure includes a nanosheet 1b' of the PG transistor and a nanosheet 2b' of the middle transistor. These nanosheets 1b' and 2b' are separated by MDI 3. The nanosheet structure is exemplarily formed on the substrate layer using a shallow STI oxide 5. The upper nanosheet 2b' of the middle transistor is in contact with the top SD Epi structure 10, and the lower nanosheet 1b' of the PG transistor is in contact with the bottom SD Epi structure 9. The dummy gate 6, the gate hard mask 7, the gate spacer 8 and the ILD 11 are shown in FIG. Figure 2 shown.

[0091] Figure 6 (b) shows that the dummy gate 6 and the gate hard mask 7 are removed in the next process step.

[0092] Figure 7 In (a) it is shown that in the next process step, a SoC coating 12 is applied around the lower nanosheet 1b' belonging to the PG transistor and also partially around the MDI 3. The SoC coating 12 may first be formed around the entire nanosheet structure and may then be etched, for example using photolithography, to expose the top nanosheet 2b' of the middle transistor.

[0093] Figure 7 In (b) it is shown that in the next process step a trench cut of the top nanosheet 2b' belonging to the nanosheet structure of the middle transistor is performed. The trench cut may include etching and may stop on the MDI 3, wherein the trench is defined by the gate spacer 8 and the ILD 11. In this case, the top nanosheet 2b' forms the channel structure of the middle transistor, and the trench cut removes at least a portion of the channel structure (note that some portions of the top nanosheet 2b' may remain under the gate spacer 8). A similar process may be used in the case where the channel structure is formed by a fin.

[0094] Figure 7 In a next process step shown in (c), the SoC coating 12 is then stripped (removed).

[0095] because Figure 5 The SRAM cell 500 can be manufactured by CFET technology, so the stacked pair consisting of the PU transistor in the first level 101 and the PD transistor in the second level 102 can be formed as a CFET. Figure 5The SRAM cell 500 may be manufactured by forming each PD transistor in the second level 102 with a nanosheet 2b having the same nanosheet width as the nanosheet 1b of the PU transistor in the first level 101, and then reducing the nanosheet width of the nanosheet 2b of each PD transistor to a second nanosheet width.

[0096] This can be done by isotropic trimming. Specifically, Figure 5 The SRAM cell 500 is processed using photolithography at the RMG process step of the manufacturing process of the SRAM cell 500. In this regard, Figure 8 and 9 shows that it can be used for processing Figure 5 FIG. 5 is an exemplary process flow for the PU and PD transistors of the SRAM cell 500 .

[0097] Figure 8 In (a) the nanosheet structure in the middle SRAM cell is shown (as it may be in the manufacturing process of the SRAM cell 500 after the CMP step of the ILD 11). Two different (vertical) cross-sectional views of the middle SRAM cell are shown on the left and right. The nanosheet structure is associated with a PU transistor in the first (lower) level 101 and a PD transistor in the second (upper) level 102. The nanosheet structure includes nanosheet 1b of the PU transistor and nanosheet 2b of the PD transistor. Similar to Figure 6 The nanosheets 1b and 2b are separated by MDI 3. The nanosheet structure is also exemplarily formed on the substrate layer using shallow STI oxide 5. The upper nanosheet 2b of the PD transistor is in contact with the top SD Epi structure 10, and the lower nanosheet 1b of the PU transistor is in contact with the bottom SD Epi structure 9. The dummy gate 6, gate hard mask 7, gate spacer 8 and ILD 11 are shown in FIG. Figure 6 shown.

[0098] Figure 8 (b) shows that the dummy gate 6 and the gate hard mask 7 are removed in the next process step.

[0099] Figure 8 In (a) it is shown that in the next process step, a SoC coating 12 is applied around the lower nanosheet 1b belonging to the PU transistor and also partially around the MDI 3. The SoC coating 12 may first be formed around the entire nanosheet structure and may then be etched, for example using photolithography, to expose the top nanosheet 2b of the PD transistor.

[0100] It is worth noting that Figure 8 Steps (a)-(c) can be done with Figure 6 In (a), (b) and Figure 7The steps in (a) are performed simultaneously.

[0101] Fig. 9 In (a) it is shown that in the next process step, an isotropic trimming of the nanosheet 2b belonging to the PD transistor is performed. The isotropic trimming may include isotropic etching from all sides, wherein the etching may not affect the MDI 3. The etching of the isotropic trimming may be performed in the opening of the SoC coating 12.

[0102] Fig. 9 In the next process step shown in (b), the SoC 12 is then stripped (removed).

[0103] Fig.10 Along Figure 5 The cross-sectional view of the cuts between A-A', B-B' and CC' shown in FIG. 5 illustrates an SRAM cell 500 for a nanosheet CFET SRAM on a HP nanosheet. It is schematically shown that the ratio of the first nanosheet width of the nanosheet 1b of the PU transistor to the third nanosheet width of the nanosheet 1b' of the PG transistor and the second nanosheet width of the nanosheet 2b of the PD transistor is 2:2:1. The nanosheet 1b of the PU transistor in the first level 101 and the nanosheet 2b of the PD transistor in the second level 102 have different widths accordingly. It can be seen how the PD transistor is arranged directly above the PU transistor and how the PG transistor is arranged next to the PU transistor in the first level 101.

[0104] Furthermore, a first level 101 including PU transistors and PG transistors is arranged on a (e.g. silicon) substrate layer. A backside supply voltage VDD is provided below the substrate layer. The backside supply voltage may be provided by a backside metal 19 and vias 18, or may be provided by a backside power rail and corresponding power rail vias (not shown). Shallow trench isolation 5 is formed in the substrate layer.

[0105] Bit lines 28 are arranged over the second level 102, and word lines 26 are arranged over bit lines 29. Word lines 26 are connected to common gate 24 through V0 vias (MINT to M1) 30, MINT layer 22, and VINT vias (gate to MINT).

[0106] Fig.11 An exemplary SRAM cell 1100 for a LP nanosheet-on-nanosheet CFET SRAM according to the present disclosure is shown. That is, Fig.11The SRAM cell 1100 is designed for LP SRAM, based on nanosheet transistors, and can use monolithic CFET technology to implement PU and PD transistors. The SRAM cell 1100 is shown in a top view, where (a) is a view of the first level 101 and (b) is a view of the second level 102. The first level 101 is arranged below the second level 102. It is worth noting that Figure 1 and Fig.11 The same elements in FIG. 1 have the same reference numerals, are implemented identically, and are not described again with reference to the above explanation.

[0107] because Fig.11 The SRAM cell 1100 is for LP SRAM, and since the PG transistor is arranged together with the PU transistor in the first level 101, the ratio of the first nanosheet width of the nanosheet 1b of the PU transistor to the third nanosheet width of the nanosheet 1b' of the PG transistor and the second nanosheet width of the nanosheet 2b of the PD transistor is 2:1:1. That is, the third nanosheet width is equal to the second nanosheet width.

[0108] because Fig.11 The SRAM cell is used for LP SRAM, so the nanosheet width of the nanosheet 1b' of the PG transistor is smaller than the nanosheet width of the nanosheet 1b of the PU transistor arranged in the same (first) level 101. Fig.11 The SRAM cell 1100 may include forming each PG transistor in the first level 101 having the same nanosheet width as the PU transistor in the first level 101, and then reducing the nanosheet width of the nanosheet 1b' of each PG transistor to the second nanosheet width of the PD transistor in the second level 102.

[0109] This can be done by isotropic trimming. Specifically, Fig.11 The SRAM cell 1100 is processed by using trimming combined with photolithography at the RMG process step of the manufacturing process of the SRAM cell 1100 for the bottom nanosheet 1b'. In this aspect, Fig.12 and 13 shows that it can be used for processing Fig.11 FIG. 1 is an exemplary process flow for a PG transistor of the SRAM cell 1100 .

[0110] Fig.12 In (a) the nanosheet structure in the middle SRAM cell is shown (it can be Figure 7In (c) after removing the SoC coating 12). Two different (vertical) cross-sectional views of the middle SRAM cell are shown on the left and right. The nanosheet structure is related to the PG transistor in the first (lower) level 101 and includes nanosheets 1b' of the PG transistor. On top of these nanosheets 1b' is the MDI 3.

[0111] Fig.12 In the next process step shown at (b), a SoC coating 12 is also applied around the nanosheet structure belonging to the PG transistor. The SoC coating 12 may first be formed around the entire nanosheet structure and then may be etched, for example using photolithography, to expose the bottom nanosheet 1b' of the PG transistor.

[0112] Fig.13 In (a) is shown isotropic trimming of the bottom nanosheet 1b' of the PG transistor in the next process step. Isotropic trimming may include etching in an isotropic manner from all sides, where the etching may not etch the MDI 3. The etching may be constrained by the SoC coating 12.

[0113] Fig.13 In a next process step shown in (b), the SoC coating 12 is then stripped (removed).

[0114] Fig.14 An exemplary SRAM cell 1400 for a HP fork-on-fork CFET SRAM according to the present disclosure is shown. That is, Fig.14 The SRAM cell 1400 is designed for HP SRAM, is based on fork-chip transistors, and uses monolithic CFET technology to implement PU and PD transistors. The SRAM cell 1400 is shown in a top view, where (a) is a view of the first level 101, and (b) is a view of the second level 102. The first level 101 is arranged below the second level 102. It is worth noting that Figure 1 and Fig.14 The same elements in FIG. 1 have the same reference numerals, are implemented identically, and are not described again.

[0115] Since the SRAM cell 1400 is for HP SRAM, and since Fig.14 The PG transistor is arranged together with the PU transistor in the first level 101, so the ratio of the first nanosheet width of the nanosheet 1c of the PU transistor to the third nanosheet width of the nanosheet 1c' of the PG transistor and the second nanosheet width of the nanosheet 2c of the PD transistor is 2:2:1. That is, the third nanosheet width is equal to the first nanosheet width.

[0116] because Fig.14The SRAM cell 1400 uses fork-sheet transistors, and adjacent transistors in the first level 101 can be processed together by a fork-sheet process, which includes forming a dielectric wall 32 that separates the nanosheet structure into first and second nanosheet structures.

[0117] because Fig.14 The SRAM cell 1400 can be manufactured using CFET technology, so a pair consisting of a PU transistor in the first level 101 and a PD transistor in the second level 102 can be formed as a CFET. Therefore, the dielectric wall 32 also exists in the second level 102. In addition, Fig.14 The SRAM cell 1400 can therefore be manufactured by forming each PD transistor in the second level 102 having the same nanosheet width as the PU transistor in the first level 101, and then reducing the nanosheet width of the nanosheet 2c of each PD transistor to the second nanosheet width.

[0118] This can be done by isotropic trimming, where the trimming can be done from one side, i.e., single-sided trimming, because the PD transistor is a fork-sheet transistor, and the other side is defined by the dielectric wall 32. Specifically, it can be Fig.14 The SRAM cell 1400 is processed by using single-side trimming combined with photolithography at the RMG process step of the manufacturing process of the SRAM cell 1400. In this regard, Fig.15 and 16 shows that it can be used for processing Fig.14 FIG. 14 is an exemplary process flow for PU and PD transistors of the SRAM cell 1400 .

[0119] Fig.15 The fork-sheet structure in the middle SRAM cell is shown in (a) (as it can be after the CMP step of ILD 11 in the manufacturing process of SRAM cell 500). Two different (vertical) cross-sectional views of the middle SRAM cell are shown on the left and right sides. The fork-sheet structure is associated with the PU transistor and the PG transistor in the first (lower) level 101 separated by dielectric walls 32, and with the PD transistor and the middle "PG" transistor in the second (upper) level 102 separated by dielectric walls 32. The fork-sheet structure includes a nanosheet 1c of the PU transistor, a nanosheet 1c' of the PG transistor, a nanosheet 2b of the PD transistor, and a nanosheet 2b' of the middle transistor. These nanosheets 1c, 1c', 2c and 2c' are separated by MDI 3. The fork-sheet structure is exemplarily formed on the substrate layer using a shallow STI oxide 5. The upper nanosheets 2c and 2c' are in contact with the top SD Epi structure 10, and the lower nanosheets 1c and 1c' are in contact with the bottom SD Epi structure 9. The dummy gate 6, the gate hard mask 7, the gate spacer 8 and the ILD 11 are, for example, Figure 5 shown.

[0120] Fig.15 (b) shows that the dummy gate 6 and the gate hard mask 7 are removed in the next process step.

[0121] Fig.15 In (c) it is shown that in the next process step, a SoC coating 12 is applied around the lower nanosheets 1c and 1c' belonging to the PG transistor and the PU transistor and also partially around the MDI 3. The SoC coating 12 can first be formed around the entire nanosheet structure and can then be etched, for example using photolithography, to expose the top nanosheet 2c of the PD transistor.

[0122] Fig.16 In (a) is shown that in the next process step a single-sided trimming of the top nanosheet 2c of the fork sheet structure belonging to the PD transistor is performed. The trimming may include etching from one side in an isotropic manner, where the etching may not affect the MDI 3 and the SoC coating 12. The etching may therefore be constrained by the SoC coating 12.

[0123] Fig.16 In a next process step shown in (b), the SoC coating 12 is then stripped (removed).

[0124] In a further processing step (not shown), the top nanosheet 2c' of the middle ("PG") transistor may be removed, for example by a similar single-sided trimming step.

[0125] Fig.17 An SRAM cell 1700 for a LP fork-on-chip CFET SRAM is shown. That is, Fig.17 The SRAM cell 1700 is designed for LP SRAM, based on fork-chip transistors, and can use monolithic CFETs to implement PU and PD transistors. The SRAM cell 1700 is shown in a top view, where (a) is a view of the first level 101 and (b) is a view of the second level 102. The first level 101 is arranged below the second level 102. It is worth noting that Figure 1 and Fig.17 The same elements in FIG. 1 have the same reference numerals, are implemented identically, and are not described again with reference to the above explanation.

[0126] Since the SRAM cell 1700 is for LP SRAM, and since the PG transistor is arranged together with the PU transistor in the first level 101, the ratio of the first nanosheet width of the nanosheet 1c of the PU transistor to the third nanosheet width of the nanosheet 1c' of the PG transistor and the second nanosheet width of the nanosheet 2c of the PD transistor is 2:1:1. That is, the third nanosheet width is equal to the second nanosheet width.

[0127] Fig.18 A method 180 according to the present disclosure is shown, which can generally be used to manufacture any of the SRAM cells 100, 400, 500, 1100, 1400, and 1700 described in the present disclosure. The method 180 includes a step 181 of forming two PU transistors in the first level 101, and a step 182 of forming two PD transistors in the second level 102. In addition, the method 180 includes a step 183 of forming two PG transistors in the first level 101 or the second level 102. The method 180 also includes a step 184 of connecting the two PU transistors and the two PD transistors to form a pair of cross-coupled inverters.

[0128] Method 180 may form each transistor as a fin transistor. In this case, each PU transistor is formed with a first number of fins 1a, each PD transistor is formed with a second number of fins 2a, and if the PG transistors (with fins 1a') are arranged in the first level 101, the ratio of the first number to the second number is 2:1, or if the PG transistors are arranged in the second level 102, the ratio of the first number to the second number is 1:2.

[0129] Method 180 may also form each transistor as a nanosheet-based transistor. In this case, each PU transistor is formed as a nanosheet 1b or 1c having a first nanosheet width, each PD transistor is formed as a nanosheet 2b or 2c having a second nanosheet width, and the ratio of the first nanosheet width to the second nanosheet width is 2:1 if the PG transistor (having nanosheet 2b' or 2c') is arranged in the first level 101, or the ratio of the first nanosheet width to the second nanosheet width is 1:2 if the PG transistor is arranged in the second level 102.

[0130] The design layouts of the SRAM cells 100, 400, 500, 1100, 1400, and 1700 described in the present disclosure are beneficial because their adaptation to LP and HP can be achieved by changing the number of fins or nanosheet width to change the driver strength. This can be achieved by an efficient method 180 and process flow. In contrast, changing the fin height or the number of nanosheet stacks at the same nanosheet width to achieve different driver strengths would be more challenging or even impossible due to simultaneous logic manufacturing.

[0131] In the claims and in the description of the present disclosure, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single element may fulfill the functions of several entities or items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used in an advantageous implementation.

Claims

1. A three-dimensional 3D static random access memory SRAM cell (100, 400, 500, 1100, 1400, 1700), include: Two pull-up PU transistors arranged in a first level (101) of the SRAM cell (100, 400, 500, 1100, 1400, 1700); two pull-down PD transistors arranged in a second level (102) of the SRAM cell (100, 400, 500, 1100, 1400, 1700), the second level (102) being arranged above or below the first level (101); wherein the two PU transistors and the two PD transistors form a pair of cross-coupled inverters; and Two pass gate PG transistors arranged in the first level (101) or the second level (102); Wherein in the SRAM cell (100, 400, 500, 1100, 1400, 1700): Each transistor is a fin transistor, each PU transistor has a first number of fins (1a), each PD transistor has a second number of fins (2b), and if the PG transistors are arranged in the first level (101), the ratio of the first number to the second number is 2:1, or if the PG transistors are arranged in the second level (102), the ratio of the first number to the second number is 1:2; or Each transistor is a nanosheet-based transistor, each PU transistor has a first nanosheet (1b, 1c) width, each PD transistor has a second nanosheet (2b, 2c) width, and if the PG transistor is arranged in the first level (101), the ratio of the first nanosheet width to the second nanosheet width is 2:1, or if the PG transistor is arranged in the second level (102), the ratio of the first nanosheet width to the second nanosheet width is 1:

2.

2. The 3D SRAM cell (100, 400, 500, 1100, 1400, 1700) according to claim 1, Features: Each PG transistor has a third number of fins (1a'), and the third number is equal to the first number or the second number; or Each PG transistor has a third nanosheet (1b', 1c') width, and the third nanosheet width is equal to the first nanosheet width or the second nanosheet width.

3. The 3D SRAM cell (100, 500, 1400) according to claim 2, It is characterized in that The SRAM cell (100, 500, 1400) is used for a high performance HP SRAM, and: If the PG transistors are arranged in the first level (101), the ratio of the first number to the third number and the second number is 2:2:1, or if the PG transistors are arranged in the second level (102), the ratio of the first number to the third number and the second number is 1:2:2; or If the PG transistor is arranged in the first level (101), the ratio of the first nanosheet width to the third nanosheet width and the second nanosheet width is 2:2:1, or if the PG transistor is arranged in the second level (102), the ratio of the first nanosheet width to the third nanosheet width and the second nanosheet width is 1:2:

2.

4. The 3D SRAM cell (400, 1100, 1700) according to claim 2, It is characterized in that The SRAM cell (400, 1100, 1700) is used for low power LP SRAM, and: If the PG transistors are arranged in the first level (101), the ratio of the first number to the third number and the second number is 2:1:1, or if the PG transistors are arranged in the second level (102), the ratio of the first number to the third number and the second number is 1:1:2; or If the PG transistor is arranged in the first level (101), the ratio of the first nanosheet width to the third nanosheet width and the second nanosheet width is 2:1:1, or if the PG transistor is arranged in the second level (102), the ratio of the first nanosheet width to the third nanosheet width and the second nanosheet width is 1:1:

2.

5. The 3D SRAM cell (100, 400, 500, 1100, 1400, 1700) according to one of claims 1 to 4, Features: Each PD transistor in the second level (102) is stacked directly above or directly below one of the PU transistors in the first level (101); and / or Each pair of transistors consisting of a PU transistor in the first level (101) and a PD transistor in the second level (102) is based on a complementary field effect transistor CFET.

6. The 3D SRAM cell (100, 400, 500, 1100, 1400, 1700) according to one of claims 1 to 5, Features: If the PG transistors are arranged in the first level (101), the PG transistors and the PU transistors in the first level (101) are p-type metal oxide semiconductor (PMOS) transistors, and the PD transistors in the second level (102) are n-type metal oxide semiconductor (NMOS) transistors; or If the PG transistor is arranged in the second level (102), the PG transistor and the PD transistor in the second level (102) are n-type metal oxide semiconductor (NMOS) transistors, and the PU transistor in the first level (101) is a p-type metal oxide semiconductor (PMOS) transistor.

7. The 3D SRAM cell (100, 400, 500, 1100, 1400, 1700) according to one of claims 1 to 6, It is characterized in that A cross-coupling structure for a pair of cross-coupled inverters is arranged in the second level (102) just above or just below each PG transistor.

8. A method (180) for processing a transistor of a three-dimensional static random access SRAM cell (100, 400, 500, 1100, 1400, 1700), the method (180) include: forming (181) two pull-up PU transistors in a first level (101) of the SRAM cell (100, 400, 500, 1100, 1400, 1700); forming (182) two pull-down PD transistors in a second level (102) of the SRAM cell (100, 400, 500, 1100, 1400, 1700), the second level (102) being formed above the first level (101), or the first level (101) being formed above the second level (102); forming two pass gate PG transistors in the first level (101) or the second level (102); connecting the two PU transistors and the two PD transistors to form a pair of cross-coupled inverters; and in: Each transistor is a fin transistor, each PU transistor is formed to have a first number of fins (1a), each PD transistor is formed to have a second number of fins (2a), and if the PG transistor is arranged in the first level (101), the ratio of the first number to the second number is 2:1, or if the PG transistor is arranged in the second level (102), the ratio of the first number to the second number is 1:2; or Each transistor is a nanosheet-based transistor, each PU transistor is formed to have a first nanosheet (1b, 1c) width, each PD transistor is formed to have a second nanosheet (2b, 2c) width, and if the PG transistor is arranged in the first level (101), the ratio of the first nanosheet width to the second nanosheet width is 2:1, or if the PG transistor is arranged in the second level (102), the ratio of the first nanosheet width to the second nanosheet width is 1:

2.

9. The method (180) according to claim 8, Features: Each PG transistor is formed to have a third number of fins (1a'), and the third number is equal to the first number or the second number; or Each PG transistor is formed to have a third nanosheet (1b', 1c') width, and the third nanosheet width is equal to the first nanosheet width or the second nanosheet width.

10. The method (180) according to claim 8 or 9, It is characterized in that Also includes: forming two intermediate transistors (2c') in the second level (102) or the first level (101), respectively, directly above or directly below the PG transistor in the first level (101) or the second level (102); removing at least a portion of the channel structure of each intermediate transistor (2c'); as well as In the space created by removing the at least a portion of the channel structure of each intermediate transistor (2c'), a cross-coupled structure for the pair of cross-coupled inverters is formed.

11. The method (180) according to one of claims 8 to 10, It is characterized in that All transistors are fin transistors, and the method (180) includes: forming each PD transistor in the second level (102) with the same number of fins (2a) as the PU transistor in the first level (101); and the method (180) further comprising: If PG transistors are arranged in the first level (101), at least one fin (2a) of each PD transistor is removed so that the number of fins (2a) of the PD transistors is reduced to the second number; or If PG transistors are arranged in the second level (102), at least one fin (1a) of each PU transistor is removed so that the number of fins (1a) of the PU transistors is reduced to the first number.

12. The method (180) according to claim 11, It is characterized in that Also includes: Each PG transistor is initially formed in the first level (101) with the same number of fins (1a') as a PU transistor, or each PG transistor is initially formed in the second level (102) with the same number of fins (1a') as a PD transistor; and the method further comprises: If PG transistors are formed in the first level (101), removing at least one fin (1a') of each PG transistor so as to reduce the number of fins (1a') of the PG transistors to the second number; or If PG transistors are formed in the second level (102), at least one fin (1a') of each PG transistor is removed so that the number of fins (1a') of the PG transistors is reduced to the first number.

13. The method (180) according to one of claims 8 to 10, It is characterized in that All transistors are nanosheet-based transistors, and the method (180) comprises: Initially forming each PD transistor in the second level (102) with a nanosheet (2b, 2c) having the same width as the PU transistor in the first level (101); and the method (180) further comprising: If PG transistors are arranged in the first level (101), reducing the nanosheet (2b, 2c) width of each PD transistor to the second nanosheet width by isotropic trimming; or If PG transistors are arranged in the second level (102), the nanosheet (2b) width of each PU transistor is reduced to the first nanosheet width by isotropic trimming.

14. The method (180) according to claim 13, It is characterized in that Also includes: Forming each PG transistor in the first level (101) with a nanosheet (1b', 1c') having the same width as the PU transistor in the first level (101); and the method further comprises: If PG transistors are formed in the first level (101), reducing the nanosheet width (1b', 1c') of each PG transistor to the second nanosheet width by isotropic trimming; or If PG transistors are formed in the second level (102), the nanosheet (1b', 1c') width of each PG transistor is reduced to the first nanosheet width by isotropic trimming.

15. The method (180) as claimed in one of claims 10 to 14, Features: The number of fins (2a) of the PD transistor is reduced, or the PD crystal is isotropically trimmed to reduce The width of the nanosheet (2b, 2c) of the body tube is in the SRAM cell (100, 400, 500, 1100, 1400, 1700) is performed at the replacement metal gate RMG process step of the manufacturing process.