Abstract:
Techniques are disclosed for reducing off-state leakage of fin-based transistors through the use of a sub-fin passivation layer. In some cases, the techniques include forming sacrificial fins in a bulk silicon substrate and depositing and planarizing shallow trench isolation (STI) material, removing and replacing the sacrificial silicon fins with a replacement material (e.g., SiGe or III-V material), removing at least a portion of the STI material to expose the sub-fin areas of the replacement fins, applying a passivating layer/treatment/agent to the exposed sub-fins, and re-depositing and planarizing additional STI material. Standard transistor forming processes can then be carried out to complete the transistor device. The techniques generally provide the ability to add arbitrary passivation layers for structures that are grown in STI-based trenches. The passivation layer inhibits sub-fin source-to-drain (and drain-to-source) current leakage.
Abstract:
Techniques are disclosed for transistor fabrication including a sacrificial protective layer for source/drain (S/D) regions to minimize contact resistance. The sacrificial protective layer may be selectively deposited on S/D regions after such regions have been formed, but prior to the deposition of an insulator layer on the S/D regions. Subsequently, after contact trench etch is performed, an additional etch process may be performed to remove the sacrificial protective layer and expose a clean S/D surface. Thus, the sacrificial protective layer can protect the contact locations of the S/D regions from contamination (e.g., oxidation or nitridation) caused by insulator layer deposition. The sacrificial protective layer can also protect the S/D regions from undesired insulator material remaining on the S/D contact surface, particularly for non-planar transistor structures (e.g., finned or nanowire/nanoribbon transistor structures).
Abstract:
Techniques are disclosed for forming transistor devices having reduced parasitic contact resistance relative to conventional devices. The techniques can be implemented, for example, using a standard contact stack such as a series of metals on, for example, silicon or silicon germanium (SiGe) source/drain regions. In accordance with one example such embodiment, an intermediate boron doped germanium layer is provided between the source/drain and contact metals to significantly reduce contact resistance. Numerous transistor configurations and suitable fabrication processes will be apparent in light of this disclosure, including both planar and non-planar transistor structures (e.g., FinFETs), as well as strained and unstrained channel structures. Graded buffering can be used to reduce misfit dislocation. The techniques are particularly well-suited for implementing p-type devices, but can be used for n-type devices if so desired.
Abstract:
Techniques are disclosed for forming low contact resistance transistor devices. A p-type germanium layer is provided between p-type source/drain regions and their respective contact metals, and an n-type III-V semiconductor material layer is provided between n-type source/drain regions and their respective contact metals. The n-type III-V semiconductor material layer may have a small bandgap (e.g.,
Abstract:
Techniques are disclosed for customization of nanowire transistor devices to provide a diverse range of channel configurations and/or material systems within the same integrated circuit die. In accordance with one example embodiment, sacrificial fins are removed and replaced with custom material stacks of arbitrary composition and strain suitable for a given application. In one such case, each of a first set of the sacrificial fins is recessed or otherwise removed and replaced with a p-type layer stack, and each of a second set of the sacrificial fins is recessed or otherwise removed and replaced with an n-type layer stack. The p-type layer stack can be completely independent of the process for the n-type layer stack, and vice-versa. Numerous other circuit configurations and device variations are enabled using the techniques provided herein.
Abstract:
Techniques are disclosed for forming transistor devices having reduced parasitic contact resistance relative to conventional devices. The techniques can be implemented, for example, using a metal contact such as one or more metals/alloys on silicon or silicon germanium (SiGe) source/drain regions. In accordance with one example embodiment, an intermediate tin doped III-V material layer is provided between the source/drain and contact metal to significantly reduce contact resistance. Partial or complete oxidation of the tin doped layer can be used to further improve contact resistance. In some example cases, the tin doped III-V material layer has a semiconducting phase near the substrate and an oxide phase near the metal contact. Numerous transistor configurations and suitable fabrication processes will be apparent in light of this disclosure, including both planar and non-planar transistor structures (e.g., FinFETs, nanowire transistors, etc), as well as strained and unstrained channel structures.
Abstract:
Techniques are disclosed for performing silicon (Si) substrate modification to enable formation of a thin, relaxed germanium (Ge)-based layer on the modified Si substrate. The thin, relaxed, Ge-based layer (e.g., having a thickness of at most 500 nm) can then serve as a template for the growth of compressively strained PMOS channel material and tensile strained NMOS channel material to achieve gains in hole and electron mobility, respectively, in the channel regions of the devices. Such a relatively thin Ge-based layer can be formed with suitable surface quality/relaxation levels due to the modification of the Si substrate, where such modification may include depositing a modification layer or performing ion implantation in/on the Si substrate. The modification layer can be characterized by the nucleation of defects which predominantly terminate within the Si substrate or the Ge-based layer, rather than running through to the top of the Ge-based layer.
Abstract:
Techniques are disclosed for forming germanium (Ge)-rich channel transistors including one or more dopant diffusion barrier elements. The introduction of one or more dopant diffusion elements into at least a portion of a given source/drain (S/D) region helps inhibit the undesired diffusion of dopant (e.g., B, P, or As) into the adjacent Ge-rich channel region. In some embodiments, the elements that may be included in a given S/D region to help prevent the undesired dopant diffusion include at least one of tin and relatively high silicon. Further, in some such embodiments, carbon may also be included to help prevent the undesired dopant diffusion. In some embodiments, the one or more dopant diffusion barrier elements may be included in an interfacial layer between a given S/D region and the Ge-rich channel region and/or throughout at least a majority of a given S/D region. Numerous embodiments, configurations, and variations will be apparent.
Abstract:
Integrated circuit transistor structures are disclosed that include a single crystal buffer structure that is lattice matched to the underlying single crystal silicon substrate. The buffer structure may be used to reduce sub-fin leakage in non-planar transistors, but can also be used in planar configurations. In some embodiments, the buffer structure is a single continuous layer of high bandgap dielectric material that is lattice matched to silicon. The techniques below can be utilized on NMOS and PMOS transistors, including any number of group IV and III-V semiconductor channel materials.
Abstract:
Techniques are disclosed for incorporating high mobility strained channels into fin-based NMOS transistors (e.g., FinFETs such as double-gate, trigate, etc), wherein a stress material is cladded onto the channel area of the fin. In one example embodiment, a germanium or silicon germanium film is cladded onto silicon fins in order to provide a desired tensile strain in the core of the fin, although other fin and cladding materials can be used. The techniques are compatible with typical process flows, and cladding deposition can occur at a plurality of locations within typical process flow. In various embodiments, fins may be formed with a minimum width (or later thinned) so as to improve transistor performance. In some embodiments, a thinned fin also increases tensile strain across the core of a cladded fin. In some cases, strain in the core may be further enhanced by adding an embedded silicon epitaxial source and drain.