Method for reducing critical dimension of patterns generated using a lithography tool

[0001] The invention relates to semiconductor device fabrication and more particularly to forming patterned layers on a semiconductor substrate.

[0002] In general, semiconductor device fabrication achieves high densities of transistors on a substrate by using extreme ultraviolet (EUV) lithography (e.g., 10-121 nm lithography) to form structures having small critical dimensions (i.e., CDs, e.g., line widths or feature widths). For example, semiconductor device fabrication processes that define minimum polysilicon lengths, active widths, or metal length of 28 nm or smaller use EUV lithography to define those CD patterns. However, EUV lithography tools are expensive and are in limited supply. Accordingly, other techniques for forming semiconductor structures having small CDs on a substrate to achieve a high transistor density are desired.

[0003] In at least one embodiment, a method for manufacturing an integrated circuit includes forming a first hard mask on a substrate by patterning a first layer of a first material. The first hard mask has first features with a first critical dimension. The method includes forming a second hard mask of a second material on the substrate by chemically modifying the first hard mask. The second hard mask has second features with a second critical dimension. The second critical dimension is smaller than the first critical dimension. The method includes forming a third hard mask by patterning a second layer of the first material on the substrate. The third hard mask has third features with the second critical dimension. The method includes forming a fourth hard mask of the second material on the substrate by chemically modifying the third hard mask. The fourth hard mask has fourth features with a third critical dimension. The third critical dimension is smaller than the second critical dimension.

[0004] In at least one embodiment, a method for manufacturing an integrated circuit includes forming a first hard mask by lithographically patterning a first sacrificial layer formed on a substrate. The first hard mask has first features with a first critical dimension. The method includes serially forming additional hard masks by chemically modifying the first hard mask and a first intermediate hard mask of the additional hard masks. A final hard mask of the additional hard masks has a second critical dimension. The second critical dimension is smaller than one half the first critical dimension.

[0005] In at least one embodiment, a method for manufacturing an integrated circuit includes using a lithographic process defined by a first minimum feature size of a first critical dimension to form a first hard mask from a first sacrificial layer on a substrate. The method includes generating a final hard mask using the first hard mask. The final hard mask has a second critical dimension. The second critical dimension is smaller than one half the first critical dimension. The final hard mask may be generated by serially reducing corresponding minimum feature sizes by chemical modification of the first sacrificial layer formed on the substrate and an additional sacrificial layer formed on the substrate.

[0009] The use of the same reference symbols in different drawings indicates similar or identical items.

[0010] A technique forms semiconductor structures having CDs smaller than the minimum lithographic CD achievable by an available lithography tool used to form structures on a substrate. The technique forms structures having that minimum lithographic CD and reduces the CD of those structures using non-lithographic techniques. The available lithography tool may be a lower-end model, and thus less expensive and more abundant, than a state-of-the-art lithography tool and forms structures having a minimum lithographic CD that is greater than twice the minimum lithographic CD of a current state of the art lithography tool.

[0011] In at least one embodiment, a method for manufacturing a semiconductor device uses an available lithography tool having a minimum lithographic CD larger than two times the CD of a final semiconductor feature. The technique uses polysilicon as a hard mask, grows silicon dioxide (i.e., oxide) using the polysilicon hard mask to shrink the length of the polysilicon, and then further reduces the length of the polysilicon. For example, an available lithography tool having a minimum lithographic CD of 56 nm is used to generate semiconductor structures having CDs of approximately 14 nm. The technique can be applied using an available lithography tool having a minimum lithographic CD of another width to generate semiconductor structures (e.g., polysilicon length, active areas, contact and metal CD, or other semiconductor structures) having CDs smaller than one half the width of the other minimum lithographic CD (e.g., an available lithography tool used to define 90 nm, 40 nm, or 28 nm polysilicon lengths can be used to generate polysilicon lengths of 22 nm, 10 nm, or 7 nm, respectively).

[0012] Referring to FIGS. 1A and 2A, in at least one embodiment, an available lithography tool is used to form a polysilicon hard mask on a substrate. In an embodiment, the available lithography tool is a lower end tool than a state-of-the-art lithography tool. The substrate has already been processed to include etch and chemical mechanical stop layers, or other appropriate sacrificial layers, above a polysilicon layer, which is above a gate oxide layer. The polysilicon layer and gate oxide layer are used to form a gate electrode or other structure of an integrated circuit device (102). In other embodiments, other sacrificial layers may be used. In at least one embodiment, substrate 202 is a semiconductor substrate (e.g., a substrate formed from silicon, germanium, other semiconducting material, or a compound semiconductor material such as gallium-arsenide) and includes multiple layers that are formed on substrate 202. In at least one embodiment, gate oxide layer 204, polysilicon layer 206, buffer oxide layer 208, silicon nitride layer 210, and buffer oxide layer 212 are formed on substrate 202. In addition, a first sacrificial polysilicon layer is patterned using the available lithography tool to form polysilicon hard mask 214 on the layers on substrate 202.

[0013] In an embodiment, gate oxide layer 204 has a thickness in a range of 2-20 nm and is formed by a conventional semiconductor manufacturing technique, e.g., thermal oxidation of a silicon substrate. Polysilicon layer 206 is a layer of polysilicon having a thickness in a range of 50-150 nm, and is formed by a conventional semiconductor manufacturing technique, e.g., using a furnace deposition process. In an embodiment, buffer oxide layer 208, is a layer of tetraethyl orthosilicate (TEOS) that is formed by low pressure chemical vapor deposition (LPCVD) and has a thickness in a range of 20-100 nm. Silicon nitride layer 210, having a thickness in a range of 20-100 nm, is a layer of silicon nitride (i.e., Si₃N₄) formed by a furnace deposition process. In an embodiment, buffer oxide layer 212 is a TEOS layer having a thickness in a range of 50-150 nm and is formed by LPCVD. Note that other embodiments may use other materials having different thicknesses and formed by other semiconductor processing techniques.

[0014] In an embodiment, polysilicon hard mask 214 is formed by first forming a polysilicon layer having a thickness in a range of 50-150 nm and using the available lithography tool to pattern the polysilicon layer to form the polysilicon hard mask. For example, a photoresist layer is formed on the polysilicon layer, e.g., by spin coating the wafer and prebaking the coated wafer. The available lithography tool exposes the coated wafer to light that is projected onto the coated wafer through a photomask and ultra-high-performance lenses to form a pattern on the coated wafer. Following a post-exposure bake and development, polysilicon hard mask 214 including void 216 remains on the substrate. In at least one embodiment, the polysilicon hard mask has a CD of width d1, e.g., equal to 56 nm, for lines and a width of d2, e.g., 56 nm, for voids between the lines. In at least one embodiment, width d1 is equal to width d2, although in other embodiments, widths d1 and d2 have different dimensions.

[0015] Referring to FIGS. 1A and 2B, the polysilicon hard mask is used to grow oxide (104). For example, the polysilicon hard mask is a patterned polysilicon layer that originally includes structures of a first composition having two sidewalls and a top surface. The sidewall and top surfaces are exposed to a chemical environment that chemically modifies the exposed surfaces of the sidewall and the exposed top surfaces to a predetermined depth, thereby creating chemically modified sidewalls and a top portion of another composition. Growth of an oxide of thickness d consumes a layer of polysilicon with a thickness of 0.46 d. In at least one embodiment, consumption of 1 Å of polysilicon grows approximately 2 Å (e.g., 2.17 Å) of oxide. In at least one embodiment, the chemical environment is an oxygen-containing environment, e.g., an environment containing molecular oxygen, carbon monoxide, carbon dioxide, or nitrogen dioxide. In at least one embodiment, the chemical environment includes several percent of hydrochloric acid, hydrogen chloride, or trichloroethylene. The chlorine removes metal ions that may occur in the oxide. The time required to grow an oxide of a particular thickness at a constant temperature on a bare silicon surface may be determined using the Deal-Grove model or the Massoud model known in the art. In at least one embodiment, wet oxidation (e.g., oxidation using water vapor), which has a higher growth rate, is used. In at least one embodiment, a combination of wet and dry oxidation is used, e.g., a dry-wet-dry cycle.

[0016] The structures after oxidation include oxide sidewalls 222 and 226, each having width d4, oxide cap 224, and polysilicon mandrel 220 having width d3. In an embodiment, width d3 (e.g., 28 nm) is approximately equal to width d4 (e.g., 28 nm). In other embodiments, width d3 is different from width d4. Sidewall 226 is spaced by width d5 from a next adjacent sidewall. In an embodiment, width d5 (e.g., 28 nm) is approximately equal to width d4, but in other embodiments, width d5 is different from width d4. Next, the top oxide, including oxide cap 224 is removed while leaving oxide sidewalls 222 and 226 (106). The top oxide may be removed using a dry etching process or other suitable technique (e.g., chemical mechanical planarization or polishing (CMP)). In at least one embodiment, the etching processes uses plasma created with oxygen, argon, carbon tetrafluoride, or other appropriate gases. FIG. 2C illustrates the resulting structure, which includes oxide sidewalls 222 and 226 (i.e., chemically modified portions) and polysilicon mandrel 220 (i.e., a nonmodified portion) having an exposed top surface.

[0017] Referring to FIGS. 1A and 2C, polysilicon mandrel 220 is selectively etched down to buffer oxide layer 212 using a selective etching process to leave behind oxide sidewalls 222 and 226 (i.e., chemically modified portions) (108). In at least one embodiment, selective etching agents that etch polysilicon and do not substantially etch oxide are used. In a conventional dry etch process, the etch rate of polysilicon is substantially greater than the etch rate of oxide, e.g., greater than 100:1, thus only a negligible amount of oxide is lost. After the selective etch, sidewalls 222 and 226 form a second hard mask, oxide hard mask 228, as illustrated in FIG. 2D. Referring to FIGS. 1A and 2D, next, oxide hard mask 228 and a conventional semiconductor manufacturing technique (e.g., dry etch) are used to etch into buffer oxide layer 212 down to silicon nitride layer 210 (110). The dry etch also removes the oxide hard mask, thereby leaving only patterned buffer oxide layer 230 above silicon nitride layer 210 (112), as illustrated in FIG. 2E.

[0018] Referring to FIGS. 1A and 2F, a conventional semiconductor manufacturing technique, e.g., a furnace deposition process, forms a second sacrificial polysilicon layer having a thickness in a range of 50-150 nm and conforming to structures already existing on the substrate (114). Sacrificial polysilicon layer 232 fills in the voids in patterned buffer oxide layer 230. A planarization technique (e.g., CMP, or other suitable technique) removes top portions of sacrificial polysilicon layer 232, thereby exposing the top surface of patterned buffer oxide layer 230 with sacrificial polysilicon structures 234 and 236 in the gaps in patterned buffer oxide layer 230 (116), as shown in FIG. 2G.

[0019] Referring to FIGS. 1A and 2H, a conventional semiconductor manufacturing technique, e.g., a selective etching process, removes patterned buffer oxide layer 230 down to silicon nitride layer 210, while preserving polysilicon hard mask 238, including structure 234 with width d3, and structure 236, having width d5, a void of width d4 between structures 234 and 236, and other sacrificial polysilicon structures that collectively form a third hard mask, polysilicon hard mask 238 (118). In a conventional dry etch process, the etch rate of oxide is substantially greater than the etch rate of polysilicon, e.g., greater than 100:1, thus only a negligible amount of polysilicon is lost.

[0020] Referring to FIGS. 1A and 2I, a second chemically modifying step is performed that exposes the sidewall and top surfaces of the third hard mask to a chemical environment. The third hard mask is chemically modified and consumes surfaces of sidewalls and top surfaces of polysilicon to a predetermined depth in the polysilicon material (120). The exposure to the chemical environment creates a chemically modified sidewall having width d6 and cap layer of another composition surrounding polysilicon mandrel 242 having width d7. Growth of an oxide thickness of d consumes a layer of polysilicon with a thickness of 0.46 d. In at least one embodiment, consumption of 1 Å of polysilicon grows approximately 2 Å (e.g., 2.17 Å) of oxide. In at least one embodiment, the chemical environment is an oxygen-containing environment, e.g., an environment containing molecular oxygen, carbon monoxide, carbon dioxide, or nitrogen dioxide. In at least one embodiment, the chemical environment includes several percent of hydrochloric acid, hydrogen chloride, or trichloroethylene. The chlorine removes metal ions that may occur in the oxide. The time required to grow an oxide of a particular thickness at a constant temperature on a bare silicon surface may be determined using the Deal-Grove model or the Massoud model known in the art. In at least one embodiment, wet oxidation, which has a higher growth rate, is used. In at least one embodiment, a combination of wet and dry oxidation is used, e.g., a dry-wet-dry cycle. In an embodiment, width d6 (e.g., 14 nm) of a sidewall is approximately equal to width d7 (e.g., 14 nm) of a corresponding mandrel, but in other embodiments, width d6 is different from width d7. Sidewall 244 is spaced by width d8 from a next adjacent sidewall. In an embodiment, width d7 (e.g., 14 nm) is approximately equal to width d8 (e.g., 14 nm), but in other embodiments, width d7 is different from width d8.

[0021] Referring to FIGS. 1B and 2I, next, the top of the oxide, including the oxide cap, is removed while leaving oxide sidewalls and the polysilicon mandrel (122). The top oxide may be removed using an etching process or other suitable technique. In at least one embodiment, the etching process uses plasma created with oxygen, argon, carbon tetrafluoride, or other appropriate gases. The resulting structure includes oxide sidewalls 240 and 244 (i.e., chemically modified portions) and polysilicon mandrel 242 (i.e., a nonmodified portion) having an exposed top, as illustrated in FIG. 2J. Next, polysilicon mandrel 242 is selectively removed using a selective etching process that etches down to silicon nitride layer 210 to leave behind the oxide sidewalls 240 and 244 (i.e., chemically modified portions) (124), as illustrated in FIGS. 1B and 2K. In at least one embodiment, etching agents that etch polysilicon and do not substantially etch silicon nitride are used (124). In a conventional dry etch process, the etch rate of silicon nitride is substantially greater than the etch rate of polysilicon, e.g., greater than 100:1, thus only a negligible amount of silicon nitride is lost. After the selective etch, sidewalls 240 and 244 form part of a fourth hard mask. Then, an etch process uses the fourth hard mask to etch through exposed portions silicon nitride layer 210 and down through buffer oxide layer 208, polysilicon layer 206, and gate oxide layer 204, to substrate 202 (126). The etch process also reduces the oxide of the fourth hard mask, but since the fourth hard mask is formed from sufficiently thick material, some material remains. The resulting patterned stacks of layers having width d6 and spaced by width d7 or width d8 are illustrated in FIG. 2L.

[0022] Referring to FIGS. 1B and 2M, the oxide, silicon nitride, and buffer oxide hard mask are removed by another removal process, thereby revealing the final polysilicon and gate oxide features on substrate 202 (128). The hard mask is removed using wet or dry processing techniques that are highly selective between oxide and silicon nitride with respect to polysilicon. The final critical dimension of polysilicon 256 and gate oxide 258 is width d6 and the polysilicon and gate oxide features are separated from adjacent polysilicon and gate oxide features by width d7 or width d8.

[0023] Although FIGS. 2A-2M describe using two iterations of forming a polysilicon hard mask and growing oxide using the polysilicon hard mask to reduce the CD of a final polysilicon and gate oxide feature, one of skill in the art will appreciate that the teachings herein can be adapted to further reduce the CD of a feature formed from different materials using additional iterations of forming a polysilicon hard mask and growing oxide using the polysilicon hard mask to reduce the CD of the feature.

[0024] Thus, techniques for forming semiconductor structures having a CD smaller than the minimum lithographic CD capability of an available lithography tool used to form structures on a substrate have been described. The technique reduces the cost of manufacturing semiconductor devices with structures having CDs at the limits of the state-of-the-art lithography tools. The technique may be used to reduce any delay from obtaining state-of-the-art lithography tools that are in short supply. The description of the invention set forth herein is illustrative and is not intended to limit the scope of the invention as set forth in the following claims. For example, while the invention has been described in an embodiment in which a gate electrode of a MOS transistor is formed, one of skill in the art will appreciate that the teachings herein can be utilized to form other features of a semiconductor device (e.g., active area, contact or metal CD, or other features). In addition, while the invention has been described in an embodiment in which structures are formed that have a CD smaller than the minimum lithographic CD capability of an available lithography tool, in other embodiments, structures are formed with other CDs by modifying the CD of the lithography steps or the thickness of the oxide grown on a polysilicon hard mask of the technique. The terms “first,” “second,” “third,” and so forth, as used in the claims, unless otherwise clear by context, distinguish between different items in the claims and do not otherwise indicate or imply any order in time, location, or quality. For example, “a first step,” and “a second step,” do not indicate or imply that the first step occurs in time before the second step. Variations and modifications of the embodiments disclosed herein may be made based on the description set forth herein, without departing from the scope of the invention as set forth in the following claims.