Method and structure for enhancing both NMOSFET and PMOSFET performance with a stressed film

[0001] The present invention relates generally to semiconductor device processing techniques, and, more particularly, to a method and structure for improving CMOS device performance and reliability by using single stress liner instead of dual stress liner.

[0002] More recently, dual stress liner (DSL) techniques have been introduced in order to provide different stresses in P-type MOSFET devices with respect to N-type MOSFET devices. For example, a nitride liner of a first type is formed over pMOSFETs of a CMOS device, while a nitride liner of a second type is formed over the nMOSFETs of the CMOS device. More specifically, it has been discovered that the application of a compressive stress in a pMOSFET channel in the direction of the electrical current improves carrier, hole, mobility therein, while the application of a tensile stress in an nMOSFET channel improves carrier, electron, mobility therein. Thus, the first type nitride liner over the pMOSFET devices is formed in a manner so as to achieve a compressive stress, while the second type nitride liner over the nMOSFET devices is formed in a manner so as to achieve a tensile stress.

[0003] For such CMOS devices employing dual liners, the conventional approach has been to form the two different nitrides using separate lithographic patterning steps. In other words, for example, the first type nitride liner is formed over both pMOSFET and nMOSFET devices, with the portions of the first type nitride liner over the nMOSFET devices being thereafter patterned and removed. After an optional formation of an oxide layer, the second type nitride liner is formed over both regions, with a second patterning step being used to subsequently remove the portions of the second type nitride liner over the pMOFET devices. Unfortunately, due to inherent inaccuracies associated with aligning lithographic levels to previous levels, the formation of the two liners could result in a gap or underlap there between. In particular, this gap will cause problems for subsequent etching of holes for metal contact vias since, during the etching, the silicide in the underlap/gap areas will be over etched. This in turn will increase sheet resistance of the silicide.

[0004] On the other hand, the two liners could also be formed in a manner such that one liner overlaps the other. In fact, the reticles used for the two separate patterning steps are typically designed to ensure an overlap such that there is no gap between the two liner materials. However, having certain regions with overlapping nitride liners creates other problems with subsequent processing due to issues such as reliability and layout inefficiencies. For example, a reactive ion etch (RIE) process for subsequent contact formation may have to accommodate for a single-thickness liner in some areas of the circuit, while also accommodating for a double-thickness (overlapping) liner in the interface areas. Moreover, if such overlapping areas are excluded from contact formation, a restriction results in terms of available layout area and critical dimension (CD) tolerances. The overlap will also cause problems during subsequent etching of holes for metal contact vias since, during the etching, all of the silicide will be over etched except for the silicide under the overlap areas. This can increase sheet resistance and junction leakage of devices.

[0005] Accordingly, it would be desirable to be able to implement the formation of a stressed CMOS device in a manner that avoids the problems discussed above related to misalignment of dual stress liners.

[0006] The foregoing discussed drawbacks and deficiencies of the prior art are overcome or alleviated by a method for forming a single stress liner for a complementary metal oxide semiconductor (CMOS) device. In an exemplary embodiment, the method includes: 1) forming a CMOS structure having an nMOSFET and pMOSFET with different gate heights (for example, the nMOSFET gate may be lower than the gate of the pMOSFET, or vice versa), 2) depositing a single stress liner of a either compressive or tensile stress over both the nMOSFET and pMOSFET; and 3) etching part of the stress liner close to the shorter of the gates to form stress of the opposite type in the channel of the shorter gate. For example, if a compressive stress liner is first formed, and the shorter gate is the nMOSFET, then etching part of the compress stress liner in proximity to the nMOSFET will result in tensile stress in the channel of the nMOSFET. If the shorter gate is the pMOSFET, then according to the invention, a tensile stress liner is deposited over both gates, and part of the stress liner is removed around the shorter pMOSFET, resulting in compressive stress in the channel of the pMOSFET.

[0011] Disclosed herein is a method and structure for improving CMOS device performance and reliability by using single stress silicon nitride liner for both nMOSFET and pMOSFET. Briefly stated, the embodiments disclosed herein result in compressive stress in the pMOSFET channel and tensile stress in the nMOSFET channel on the same chip or integrated circuit (IC) by using the same stressed film to cover both the pMOSFET and the nMOSFET. This results in performance enhancement due to local stress for both nMOSFET and pMOSFET, without causing misalignment problems.

[0012] Referring initially to FIG. 1, there is shown a cross sectional view of a semiconductor substrate 100 having an nMOSFET device region 102 and a pMOSFET device region 104 separated by an isolation region 105 formed therein, such as a shallow trench isolation (STI).

[0013] Referring to FIG. 2, a gate dielectric layer 106 is formed over the substrate 100 including the isolation region 105. The gate dielectric 106 may be any suitable dielectric material, such as silicon dioxide. The gate dielectric 106 may be formed, for example, by thermal oxidation or deposition of a high K material. The gate dielectric 106 typically has a thickness in the range of about 1–2 nm. In accordance with the invention, a first layer of a gate conductor 108 is formed atop the gate dielectric layer 106. The first gate conductor layer 108 may be any suitable gate conductor material such as polysilicon, W, Ta or SiGe, more typically polysilicon. For gate lengths of 35–45 nm, the polysilicon layer 108 is preferably 10–30 nm thick. A second gate conductor layer 110 having an etch rate different than the first gate conductor layer 108, such as polysilicon-germanium (poly-SiGe), if the first conductor layer is polysilicon, is deposited atop the first gate conductor (e.g. polysilicon) layer 108. For gate lengths of 35–45 nm, the poly-SiGe layer 110 is preferably 70–90 nm thick. Preferably, the second gate conductor layer 110 is thicker than the first conductor layer 108.

[0014] Referring to FIG. 3, devices 102, 104 are formed by processes now known or developed in the future. For example, the gate stacks may be formed by patterned etching, formation of spacers including optional thin oxide liners 112 and nitride spacers 114, and implantation to form source/drain halo regions and extensions 116, followed by source/drain anneal, as will be recognized by one skilled in the art.

[0015] Referring to FIG. 4, the pMOSFET 104 is covered by a mask such as photo resist layer 126. Then, the second gate conductor layer 110, e.g. the poly-SiGe layer, is removed from the first gate conductor layer 108 in the nMOSFET 102, for example, by an etch process selective to silicon, poly Is, oxide and nitride. Then the exposed oxide liner 112 above the first gate conductor 108 is removed from the sidewalls 114 of the nMOSFET 102, for example, using a process such as buffered HF (BHF). Etch time will depend on the thickness of the oxide liner 112. Since the oxide liner 112 is very thin, for example, on the order of about 5–10 nm, there will be no significant damage to the isolation region 105.

[0016] Referring to FIG. 5, the photo resist 126 is removed. Then, a metal layer is deposited over the structure. For example, in a preferred embodiment, nickel is deposited at a thickness between about 3–20 nm, sufficient to fully silicide the polysilicon layer 108 in the nMOSFET gate stack 102. After an anneal, for example, at 300–500 §C at 1–60 seconds, a semiconductor metal alloy is formed from the metal and the silicon of the nMOSFET gate stack 102, the silicon of the substrate 100, and the SiGe of the pMOSFET gate stack 104. The resulting structure includes silicide regions 120 over the source/drain regions 116, a fully silicided gate conductor 122 in the nMOSFET 102, and a silicided top portion 124 of the pMOSFET 104.

[0017] Next, referring to FIG. 6, the nitride spacers 114 are etched back, for example by a wet etch or dry etch process, so that the nitride spacers 114 have substantially the same height as the silicided gate conductor 122 and oxide liner 112 of the nMOSFET 102, resulting in an nMOSFET gate stack 102 that is shorter in height than the pMOSFET gate stack 104. Since a wet etch process is isotropic, the nitride spacers 114 on the pMOSFET 104 will be thinned. Preferably, the nitride spacers 114 are thinned no more than about half its original thickness.

[0018] Referring to FIG. 7, a compressive nitride film 130 is deposited over the structure. The thickness of the compressive nitride film is preferably in the range 40–100 nm. The compressive nitride material 130 may be formed by high density plasma (HDP) deposition or plasma enhanced CVD (PECVD), for example, SiH₄/NH₃/N₂ at about 200° C. to about 500° C. This results in compressive stress being generated in the channels 182, 184 of the nMOSFET and pMOSFET regions 102, 104, respectively (see FIG. 8).

[0019] Next, referring to FIG. 8, a thin etch stop layer 132, such as an oxide, for example, about 50–100 angstroms thick, is formed atop the compressive nitride layer 130. Then, a photoresist material 146 is formed over the structure and thereafter patterned so as to form openings 148 in the resist 146 that expose the surface of the thin oxide 132 on at least opposite sides of the nMOSFET 102 over the source/drain regions 116, which will be used to pattern openings 158 in the compressive nitride layer 130 (see FIG. 10). For a sufficiently narrow width device, forming the opening 158 completely around the perimeter of the gate 122 in the compressive layer 130 may enhance device performance. However, for a wide width device, the additional benefit caused by surrounding the device by openings 158 is small, and it would be sufficient to form openings 158 on opposite sides of the shorter device 102. The exposed portion of the thin oxide layer 132 above the nMOSFET device 102 is removed to form openings 151 in the thin oxide 132, using a process such as by RIE for example, stopping on the compressive nitride layer 130. Then the resist layer 146 is removed. The resulting structure is illustrated in FIG. 9.

[0020] Next, the compressive nitride layer 130 is removed, for example, by an isotropic or wet etch, where the openings 151 in the thin oxide 132 has been formed over the source/drain regions 116 of the nMOSFET device 102, to form openings 158 so that an inner edge 159 of the opening 158 is at a horizontal distance Lcut from the outer edge of the gate conductor 122, so that the stress of the channel region 182 of the nMOSFET device 102 is modified to become tensile stress. The resulting structure is illustrated in FIG. 10. It is noted that the width of the opening 158 may be from about 30 nm to about 100 nm, but is not critical, and that the edge of the opening 158 away from the gate stack may extend as far as the isolation region 105.

[0021] The preferred horizontal distance Lcut of the opening 158 from the gate conductor 122 is preferably selected so as to optimize the resulting stress in the channel region 182. This optimal distance L_Max can be determined, for example, by simulating the stress at the center 183 of the channel region 182 for a range of expected gate structures similar to that of nMOSFET device 102, but varying the Lcut distance, and then determining the position of Lcut (i.e. L_Max) to be such that the channel stress is the maximized, as illustrated in FIG. 11. For the case of a pMOSFET that is shorter than the nMOSFET, the initial stressing layer 130 is tensile, and the value of Lcut is preferably chosen at L_Max to maximize the compressive stress in the pMOSFET channel.

[0022] Next, a nitride film 162 having substantially neutral stress, or substantially without a large stress component is deposited over the structure, for example, by chemical vapor deposition (CVD) or high density plasma (HDP), so that the openings 158 are filled in the compressive nitride layer 130, as illustrated in FIG. 12. Preferably the thickness of the neutral stress layer 162 should be greater than ½ of the width of the opening 158. Then the neutral stress layer 162 is etched back to a surface that is substantially level with the surface of the thin oxide layer 132, as illustrated in FIG. 13. Subsequently, the nMOSFET device 102 and pMOSFET device 104 may be completed as known by one skilled in the art.

[0023] While the invention has been described with reference to a preferred embodiment or embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted for elements thereof without departing from the scope of the invention. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the invention without departing from the essential scope thereof. Therefore, it is intended that the invention not be limited to the particular embodiment disclosed as the best mode contemplated for carrying out this invention, but that the invention will include all embodiments falling within the scope of the appended claims.