Group III N transistors on nanoscale template structures

III-N transistor architectures with CMOS integration address the scalability and efficiency challenges of PMIC and RFIC integration, enabling high-voltage and high-frequency operations in mobile computing platforms.

DE112013005528B4Inactive Publication Date: 2024-11-07INTEL CORP
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
DE112013005528
Authority / Receiving Office
DE · DE
Patent Type
Patents
Current Assignee / Owner
Priority Date
2012-12-19
Filing Date
2013-06-24
Publication Date
2024-11-07
Estimated Expiration
Not applicable · inactive patent

Smart Images

  • Figure 00000000_0000_ABST
    Figure 00000000_0000_ABST
Patent Text Reader

Abstract

III-N field-effect transistor, hereinafter abbreviated as III-N-FET, arranged on a silicon substrate, wherein the III-N-FET comprises the following: an anchor positioned above the substrate (203), a first and a second group III N device layer stack that are physically separated from each other by the anchor, each stack having a c-axis extending in opposite directions from the anchor and parallel to a surface plane of the substrate (203), and a gate stack arranged above the III-N device layer stacks to control the conductivity of a channel semiconductor layer (217A, 217B) in each of the III-N device layer stacks, wherein the channel semiconductor layer has a conduction band offset relative to the armature material.
Need to check novelty before this filing date? Find Prior Art

Description

TECHNICAL FIELD

[0001] Embodiments of the present invention relate generally to microelectronic devices and their fabrication, and more particularly to Group III-N transistor architectures. BACKGROUND

[0002] The mobile computing device markets (e.g., smartphones and tablets) benefit from smaller component form factors and lower power consumption. Because current platform solutions for smartphones and tablets rely on multiple encapsulated integrated circuits (ICs) mounted on a printed circuit board (PCB), further scaling to smaller and more power-efficient form factors is limited. For example, a smartphone includes a separate power management IC (PMIC), radio frequency IC (RFIC), and Wi-Fi / Bluetooth / GPS IC in addition to a separate logic processor IC. System-on-a-chip (SoC) architectures offer the advantage of scaling, which cannot be achieved through component integration at the board level.Although the logic processor IC itself can be considered a system on a chip (SoC) that integrates both memory and logic functions, more comprehensive SoC solutions for mobile computing platforms have remained illusory because the PMIC and the RFIC operate at two or more of a high voltage, high power, and high frequency.

[0003] Conventional mobile computing platforms use incompatible transistor technologies specifically designed for the different functions performed by the PMIC and the RFIC. For example, the PMIC typically uses laterally diffused silicon MOS (LDMOS) technology to perform voltage conversion and power distribution (battery voltage regulation including step-up and / or step-down voltage conversion, etc.). Group III-V compound semiconductors, such as GaAs heterojunction bipolar transistors (HBTs), are typically used in the RFIC to generate sufficient power gain at GHz carrier frequencies. Conventional silicon field-effect transistors implementing CMOS technology then introduce a third transistor technology used for logic and control functions within the mobile computing platform.In addition to fundamental semiconductor material incompatibilities between the various ICs in the mobile computing platform, transistor design for DC-to-DC converter switches in the PMIC has generally been incompatible with transistor design for high-frequency power amplifiers in the RFIC. For example, the relatively low breakdown voltage of silicon requires the source-drain isolation in a DC-to-DC converter switch to be significantly larger than is permissible for a power amplifier transistor, which requires an Ft exceeding 20 GHz and potentially extending to 500 GHz, depending on the carrier frequency (for example, WPAN is 60 GHz, so transistors require an Ft many times 60 GHz). Such different transistor-level design requirements make the manufacturing processes for the various transistor designs diverse and difficult to integrate into a single process.

[0004] Therefore, although a SoC solution for the mobile computing domain that would integrate PMIC and RFIC functions is attractive to improve scalability, reduce costs, and improve the power efficiency of the platform, one barrier to a SoC solution is the lack of a scalable transistor technology that offers both sufficient speed (i.e., a sufficiently high gain cutoff frequency F t ) and a sufficiently high breakdown voltage (BV).

[0005] Group III nitride (III-N) devices offer a promising approach for integrating PMIC and RFIC functions with CMOS because they offer both high BV and high F tcan be obtained. However, heteroepitaxy of III-N material stacks on silicon substrates presents a technical challenge, at least due to significant lattice mismatch and thermal expansion mismatch, both of which can lead to high defect densities in device layers and poor device performance. Techniques and epitaxial semiconductor stack architectures that can provide reduced defect densities in device layers are therefore advantageous.

[0006] US 20070004117 A1 discloses a method for manufacturing a semiconductor device, the method comprising: forming a plurality of fins comprising a semiconductor material on an insulating layer, forming gate insulating films on the sidewalls of the fins, forming a gate electrode extending in an arrangement direction of the fins and electrically insulated from the fins, the gate electrode being arranged common to the fins on the gate insulating film, implanting an impurity into portions of the fins using the gate electrode as a mask to form a source-drain diffusion layer, the portions of the fins extending on both sides of the gate electrode, and depositing a conductive material on both sides of the fins to connect the fins to each other.

[0007] US 20120015493 A1 discloses a method for providing a high-k partial gate structure formed over a semiconductor fin. A nitride layer is formed over the gate structure and the semiconductor fin in two separate deposition processes, with the first process forming a very thin nitride film. Implantation processes may be performed between the two nitride film deposition processes. The first nitride film may be SiNx or SiCNx, and the second nitride film may be SiCNx. The nitride films may be combined to form low-wet-etch spacers, enabling further processing operations without damaging the underlying structures and without forming additional dummy spacers. Further processing operations include epitaxial silicon / SiGe processing sequences and source / drain implantation processes performed with intact low-etch spacers.

[0008] US 20130001705 A1 relates to a method for forming an integrated circuit structure. The method comprises providing a wafer comprising a substrate and a semiconductor fin on a main surface of the substrate, and performing a deposition step to epitaxially grow an epitaxial layer on a top surface and sidewalls of the semiconductor fin, wherein the epitaxial layer comprises a semiconductor material. Subsequently, an etching step is performed to remove a portion of the epitaxial layer, leaving a remaining portion of the epitaxial layer on the top surface and sidewalls of the semiconductor fin.

[0009] DE 112010002352 B4 relates to a method for producing FinFet structures with strain-inducing source / drain-forming spacers and FinFet structures. BRIEF DESCRIPTION OF THE DRAWING

[0010] Embodiments of the present invention are explained by way of example and not limitation, and can be more fully understood by reference to the following detailed description taken in conjunction with the figures.

[0011] They show: Fig. 1 is a flowchart of methods for fabricating a III-N field effect transistor (FET) according to embodiments of the invention, the Fig. 2A, Fig. 2B and Fig. 2C isometric views when operations of the procedures from Fig. 1 on a substrate, according to embodiments, the Fig. 3A, Fig. 3B and Fig. 3C Sectional views through a Fig. 2C, after certain operations of the procedures from Fig. 1 were carried out according to embodiments, the Fig. 4A, Fig. 4B and Fig. 4C shows a cross section through a channel region of a III-N FET according to an embodiment, Fig. 5 is a functional block diagram of a Group III-N SoC implementation of a mobile computing device according to an embodiment of the present invention; and Fig. 6 is a functional block diagram of a computing device according to an implementation of the invention. DETAILED DESCRIPTION

[0012] In the following description, numerous details are set forth, but it will be understood by those skilled in the art that the present invention may be practiced without these specific details. In some instances, well-known methods and devices are shown in block diagram form rather than in detail to avoid obscuring the present invention. Reference in this specification to "one embodiment" means that a particular feature, structure, function, or characteristic described in connection with the embodiment is included in at least one embodiment of the invention. Accordingly, appearances of the phrase "according to an embodiment" in various places throughout this specification are not necessarily referring to the same embodiment of the invention.Furthermore, the particular features, structures, functions, or properties may be combined as appropriate in one or more embodiments. For example, a first embodiment may be combined with a second embodiment, provided the two embodiments are not mutually exclusive.

[0013] The terms "coupled" and "connected," along with their derivatives, may be used herein to describe structural relationships between components. It should be noted that these terms are not intended to be synonymous with one another. Rather, in certain embodiments, "connected" may be used to indicate that two or more elements are in direct physical or electrical contact with one another. "Coupled" may be used to indicate that two or more elements are in either direct or indirect (with other intervening elements between them) physical or electrical contact with one another, and / or that the two or more elements cooperate or interact with one another (e.g., such as in a cause-and-effect relationship).

[0014] The terms "over," "under," "between," and "on," as used herein, refer to a relative position of a layer of material with respect to other layers. For example, a layer disposed above or below another layer may be in direct contact with the other layer or may have one or more intermediate layers. Furthermore, a layer disposed between two layers may be in direct contact with the two layers or may have one or more intermediate layers. In contrast, a first layer "on" a second layer is in direct contact with that second layer.

[0015] Described herein are embodiments of III-N MOSFETs formed on template anchors, such as silicon fin sidewalls, to achieve reduced defect densities in III-N device layers. According to embodiments, a III-N junction layer is formed on a (111) or (110) face of a compliant crystalline silicon fin sidewall. A III-N semiconductor channel is further formed on the junction layer. According to certain embodiments, the semiconductor channel comprises indium (In) to increase a conduction band offset from the template anchor material. According to other embodiments, the compliant crystalline silicon fin is sacrificial and is either removed, oxidized, or otherwise converted into a dielectric anchor during transistor fabrication after III-N epitaxy.According to certain embodiments using a sacrificial compliant epitaxial template or mandrel, the III-N semiconductor channel consists essentially of pure GaN. After removing the crystalline silicon epitaxial mandrel, the III-N transistor can support a higher breakdown voltage.

[0016] According to embodiments, the high-electron-mobility FETs described herein are used in SoC solutions that integrate an RFIC with a PMIC to implement high-voltage and / or high-power circuits. With the transistor structures described herein, SoC solutions can deliver the product-specific electrical current and power requirements needed for a mobile computing platform. The fast-switching high-voltage transistors are capable of handling high input voltage swings and providing high power efficiencies at RF frequencies. According to embodiments, the III-N transistor architectures described herein are monolithically integrated with Group IV transistor architectures, such as planar and non-planar silicon CMOS transistor technologies.According to specific embodiments, the III-N transistors described herein are used in SoC architectures that integrate high-performance wireless data transmission and / or high-voltage power management functions with low-power CMOS logic data processing. High-frequency operation suitable for broadband wireless data transmission applications is possible, while the use of III-N materials with a wide bandgap also provides a high BV value, allowing sufficient RF output power to be generated for wireless data transmission applications. This combination of a high F. t / F maxThe high-value and high-voltage capability also enable the III-N FET architectures described here to be used for high-speed switching applications in DC-DC converters that utilize reduced-size inductive elements. Because both power amplification and DC-to-DC switching applications are key functional blocks in smartphones, tablets, and other mobile platforms, the structures described here can be used in an SoC solution for such devices.

[0017] Fig. 1 is a flowchart illustrating a method 101 for fabricating a III-N field-effect transistor (FET) according to embodiments of the invention. Fig. 2A, Fig. 2B and Fig. 2C show isometric views when operations of the methods in Fig. 1 on a substrate, according to embodiments. Generally, in method 101, epitaxial growth of III-N semiconductor crystals is performed on patterned nanoscale template anchors, which are then processed during device fabrication to provide III-N FET devices of suitable functionality and which can be integrated with CMOS fabrication. At sufficiently small dimensions, the patterned epitaxial growth can remove defects in the III-N device layers from the ~1 10 9 / cm 2reduce the stresses typical of unstructured III-N growth on silicon. Because growth occurs on nanoscale surfaces of the template anchor, defects can be manipulated to propagate into the anchor structure, reducing the number of defects that would otherwise propagate into the III-N device films. One advantage of nanoscale structures is that they have a high surface-to-volume ratio, providing a large free surface area for the relaxation of stresses developed as a result of the mismatch between thermal expansion coefficients and high growth temperatures.

[0018] With reference to Fig. 1, it should be noted that the method 101 begins in operation 110 with the formation of a template structure capable of seeding and anchoring epitaxial III-N films, for example, in the form of a fin, to the substrate. According to exemplary embodiments, the template anchor structures are made of single-crystal silicon and are formed as shown in Fig. 2A, a fin 210 is formed in the substrate 203 having first and second opposing sidewalls 210A, 210B. According to the exemplary embodiment, the substrate 203 is substantially monocrystalline and is (100) silicon (i.e., having a (100) top face) or (110) silicon (i.e., having a (110) top face). For (110) silicon embodiments, the vertical sidewalls 210A, 210B are (111) faces. (111) crystal planes are advantageous for III-N epitaxial growth because the lattice mismatch is only about 16%. For (100) silicon embodiments, (110) planes are present on the fin sidewalls 210A, 210B when the sidewalls are in the <110> - direction on the (100) plane. (110) crystal planes are also advantageous for epitaxial III-N growth because (110) silicon planes have a mismatch with III-N between (111)-Si and (100)-Si (about 42%).Both (100) and (110) substrate crystal orientations are also advantageous for the formation of silicon transistors (e.g., in other regions not covered by III-N epitaxial layers) and are therefore ideal for embodiments where a group III-N transistor formed on fin 210 is to be monolithically integrated using silicon CMOS transistor technology. In particular, other substrates with similarly mismatched lattice constants may also benefit from the template anchors described herein, such as, but not limited to, substrates comprising germanium (Ge), which may be alloyed with silicon or in pure form.

[0019] According to embodiments, the epitaxial template anchor has surfaces with nanoscale dimensions. The surface area-to-volume ratio is an important property of the structured template anchor that influences the quality of the epitaxy, with a higher free surface area improving the epitaxial crystal quality. According to the Fig. In the exemplary embodiment shown in Figure 2A, the fin 210 has a critical dimension (CD) of the fin width W F of less than 50 nm and advantageously less than 20 nm, while the fin height H F is less than 100 nm and is advantageously between 25 nm and 100 nm when W F less than 20 nm. As further described herein, the III-N epitaxial layers grown on the sidewalls 210A, 210B according to the exemplary embodiment are ultimately considerably thicker than W F . The narrow fin width W FReduces the compliance of fin 210 relative to a dimensionally larger silicon block, allowing fin 210 to be loaded (e.g., compressed) to reduce stress (e.g., tensile stress) in the III-N epitaxial layers that would otherwise be induced by fin 210 if it had a larger width and were less compliant. This compliance in the seed layer of III-N epitaxy can enable much thinner III-N layers to achieve sufficient defect density for suitable III-N transistor characteristics.

[0020] According to embodiments in which the same epitaxial III-N stack is grown simultaneously on both sides of the fin, the stress (strain) is symmetrical about a longitudinal centerline of the fin 210, so that the stresses between the sides of the fin are advantageously balanced. A fin height H Fof less than 100 nm has the advantage of being on the order of magnitude of the III-N crystal grain coalescence dimension. Therefore, the fin 210 serves as a template structure with dimensions of less than 100 nm along at least two axes (z and x axes in Fig. 2). The third dimension, here associated with the fin length, can be at least an order of magnitude larger than the critical dimension axes (e.g., 1 µm or more). At such dimensions, a defect density reduction of at least three orders of magnitude can be expected relative to epitaxial III-N film growth on non-compliant seed layers (e.g., host substrate or large template structures with dimensions significantly larger than the thicknesses of the III-N epitaxial layers).

[0021] As in Fig. As further shown in Figure 2A, the fin 210 is covered with a hard mask 243, for example, silicon dioxide or another dielectric deposited by chemical vapor deposition (CVD) or another conventional technique. The hard mask 243 is patterned together with the fin 210 and may block subsequent epitaxial growth on the top surface of the fin 210. Adjacent to the fin 210, on both sides of the fin sidewalls 210A, 210B, is an isolation dielectric 241 covering the top surface of the substrate 203. The isolation dielectric 241 may be any conventional dielectric, such as those typically used for trench isolation, etc. (e.g., silicon dioxide).The presence of the isolation dielectric 241 provides isolation between epitaxial III-N layers formed on the fin sidewalls 210A, 210B, and when selective epitaxial processes are used (e.g., MOCVD), the isolation dielectric 241 may also be a beneficial means to reduce the semiconductor surface area as epitaxial growth occurs, thereby reducing microcharging effects, etc. Although the fin patterning may be performed in various ways, the example technique includes patterning the hard mask 243, recess etching the substrate 203 around the hard mask 243 (e.g., with a trench etch), depositing and planarizing the plane of the isolation dielectric 241 with the hard mask 243, and recessing the isolation dielectric 241 to expose the desired height of the fin sidewalls 210A, 210B.

[0022] To Fig. Returning to Figure 1, it should be noted that in process 115, epitaxial growth occurs on the crystalline surfaces of the fin. The fin is intended to serve both as a template / seed for growth and as a physical anchor for the substrate. With reference to Fig. 2B, it should be noted that any of the III-N semiconductor layers to be ultimately grown on the intended fin may have a c-axis that is perpendicular to the fin sidewalls 210A, 210B or substantially parallel to a top surface of the substrate 203.

[0023] According to the illustrated embodiment, a crystalline buffer or transition layer is formed epitaxially on the (111) sidewall surfaces of the template anchor as a first layer. This transition layer serves to accommodate a change in lattice constant from the template surface (e.g., silicon) to the subsequently grown III-N semiconductor channel layer. The transition layer may be comprised of one or more of a III-N material or a crystalline oxide. According to certain embodiments, the transition layer is preferably comprised of a material having a wider band gap than a channel layer subsequently grown over the III-N buffer to achieve carrier confinement and thus reduced transistor leakage current into the fin 210. For such embodiments, exemplary III-N materials include AlN, AlGaN, and GaN. In particular, for an Al x In 1-xN-layer the Al mole percentage is about 83 (for example Al 0,83 In 0,17 N), although the exact concentration can vary across the thickness of the transition layer. Although Al x In 1-x N-junction layers offer a number of advantages, the relatively low epitaxial growth temperature of Al x In 1-x N is particularly noteworthy. Regardless of whether the growth is carried out by MBE or MOCVD, MOVPE, etc., the growth of Al x In 1-x N in the range of 300 °C is lower than in many alternative III-N materials. For an AlGaN embodiment, the Al mole percentage is not greater than 30% (for example, Al <0,3 G >0,7 N), although the exact concentration may vary across the thickness of the transition layer.

[0024] Examples of crystalline dielectrics that can be used as a transition layer include crystalline wurzite nitrides, such as TiN, SiN, AlN, and crystalline wurzite oxides, such as Al2O3, Gd2O3, Sc2O3, Ta2O5, and TiO2. Such material layers are typically deposited as polycrystalline layers, and when exposed to the high growth temperature of III-N semiconductors, they tend to form wurzite crystallinity suitable for III-N growth. As further described in Fig. 2B, the transition layers 215A and 215B are formed simultaneously on the respective fin sidewalls 210A, 210B. According to advantageous embodiments, III-N transition layers 215A are formed to a thickness of less than 100 nm (where the thickness along the x-axis is Fig. 2B) are grown by MOCVD or MOVPE, while the crystalline wurcite nitrides and oxides are deposited by atomic layer deposition to a thickness of 5 to 10 nm.

[0025] To Fig. Returning to Figure 1, the method 101 then continues for the growth of the III-N channel semiconductor layer in either operation 117 or operation 118. Generally, the channel semiconductor layer is substantially single-crystalline, and although referred to herein as "monocrystalline," those of ordinary skill in the art will understand that a low level of crystal defects may nevertheless be present as artifacts of an imperfect epitaxial growth process. In general, the III-N semiconductor in the channel layer should have a relatively high carrier mobility, and therefore, according to embodiments, the channel layer consists essentially of undoped Group III nitride material (i.e., with a minimized impurity concentration) for minimal impurity scattering.

[0026] Fig. 2B further shows the III-N channel semiconductor layers 217A and 217B arranged on the transition layers 217A, 217B. According to one embodiment, in operation 117 ( Fig. 1) A III-N channel comprising indium (In) is epitaxially grown over the junction layer. By incorporating indium into the III-N channel (i.e., InGaN channel), the conduction band offset of the channel semiconductor relative to the silicon fin 210 can be significantly increased to confine charge carriers (electrons) within the channel semiconductor layer (e.g., layers 217A, 217B). In the absence of sufficient charge confinement, charge can accumulate within the template structure and impair device operation (e.g., through the formation of leaks and / or parasitic channels).Therefore, although a GaN channel offers a very small conduction band offset relative to silicon and therefore the junction layer must further serve as a back barrier or an additional back barrier must be grown over the junction layer, an InGaN channel semiconductor layer can be formed so that the overall thickness of the III-N layers grown on the epitaxial template structure can be advantageously reduced (minimized) and / or greater flexibility in the selection of buffer materials is possible. According to exemplary embodiments, the InGaN channel semiconductor layer comprises 10-20% indium, and according to certain such embodiments, the InGaN channel semiconductor layer is deposited by MOCVD or MOVPE to a thickness of at most 50 nm (x-axis in FIG. Fig. 2B) were allowed to grow up.

[0027] According to an alternative embodiment, method 101 continues in operation 118, where a GaN channel layer is grown over the junction layer. GaN is advantageous in terms of its wide band gap and the associated high breakdown voltage. For such an embodiment, the GaN channel semiconductor layer is again advantageously grown by MOCVD or MOVPE to a thickness of at most 50 nm. However, as described with reference to operation 117, carrier confinement within a GaN channel semiconductor layer is problematic if there is no junction layer with a wider band gap to serve as a barrier between the silicon template and the channel.While the transition layer may be sufficient with respect to carrier confinement according to some embodiments, according to other embodiments, for example, when the transition layer is also comprised of GaN, the presence of the silicon template anchor (e.g., fin 210) may pose a problem for carrier confinement. For such embodiments, method 101 further includes removing or material converting the epitaxial template anchor after it has fulfilled its growth seed function. The silicon fin (or a similar template anchor) may then be considered a sacrificial feature or an "epitaxial mandrel."

[0028] After process 117 or 118, a III-N polarization layer is formed in process 120 or 121, respectively. In process 120 or 121, the III-N cap or polarization layer is grown epitaxially (for example, by MOCVD or MOVPE) over the channel semiconductor layer and is intended to functionally serve as a charge-inducing layer to controllably supply charge carriers in the form of a charge layer, commonly referred to as a 2-D electron gas (2DEG, as in Fig. 4A as 219A and 219B). Fig. 2B shows exemplary polarization layers 220A, 220B, the thickness of which may be between 1 nm and 20 nm, but the thickness of which is advantageously less than 10 nm.

[0029] The polarization layer can further serve as a means for carrier confinement if the band gap is large enough. For the exemplary embodiments, the polarization layer is the source of the layer charge and serves as a top barrier to reduce and minimize the overall thickness of the epitaxial III-N material. However, according to other embodiments, a compositionally diverse charge-inducing layer may be used together with a thin top barrier layer with a different composition to enable tuning of the transistor threshold voltage while ensuring a thin (e.g., >0.5 nm) wide band gap material at the surface of the channel semiconductor layer to achieve lower alloy scattering and high carrier mobility.

[0030] Due to the different polarizations of the materials used in the III-N channel semiconductor layer and the polarization layer (or the charge-inducing layer in between), a charge density can be provided that can be controlled by selecting a work function metal as the subsequently formed gate electrode and / or controlling the semiconductor thickness along the gate length (for example, the y-dimension in Fig. 2B for the example lateral transistor and the z-dimension in Fig. 2B for a vertical transistor). The functional characteristics of the transistor depend on the materials chosen for the polarization layer, the channel semiconductor layer, and the gate electrode.

[0031] According to embodiments of method 101, either in operation 120 (where the channel layer is InGaN) or in operation 121 (where the channel layer is GaN), polarization layers 220A, 220B comprising at least one of AlInGaN, AlGaN, AlInN, or AlN are grown on the channel semiconductor layer. According to an exemplary embodiment, polarization layers 220A, 220B comprise approximately 17% In. According to embodiments, polarization layers 220A, 220B comprise only one intrinsic impurity doping level (e.g., i-Al w In 1-w N). According to other embodiments, a multilayer stack of group III nitrides (e.g., an AlInN / AlN stack, with the AlN layer of the stack adjacent to the channel semiconductor layers 217A, 217B) may be grown in either operation 120 or operation 121.

[0032] As further stated in Fig. 2B, after epitaxial growth of the III-N device layer stacks on the fin sidewalls 210A, 210B, the dielectric hard mask 243 is removed to expose the silicon fin 210 and / or planarization of the epitaxial layers is performed.

[0033] After process 121 ( Fig. 1) The method 101 continues either to operation 130, where the template anchor (e.g., silicon fin 210) is selectively removed down to the epitaxial layers (e.g., selectively down to the junction layers 215A, 215B, etc.), or to operation 135, where the template anchor (e.g., silicon fin 210) is converted into a dielectric anchor. In these two embodiments, silicon fin 210 no longer exists as a crystalline semiconductor, which is particularly advantageous where the III-N transistor is to be operated at high voltages that would otherwise induce fields in silicon fin 210, leading to silicon breakdown.Therefore, according to an embodiment advantageous for high breakdown voltage operation, wherein a GaN channel layer is disposed on a GaN junction layer, removing the silicon template serves to confine charge carriers and increase the breakdown voltage of the III-N transistor.

[0034] The channel region of a III-N FET according to embodiments of the present invention is shown in the Fig. 3A, Fig. 3B and Fig. 3C, which are sectional views corresponding to the Fig. 2C shown A-A' plane at a point before operation 160 ( Fig. 1), where the gate stack (the gate dielectric and the gate electrode) is formed on the III-N epitaxial layers. Fig. 3A corresponds to a point after process 120, and the Fig. 3B and Fig. 3C correspond to points after operations 130 and 135 respectively.

[0035] For those in Fig. 3A, the III-N epitaxial layers, as well as the silicon fin 210, are present after operation 120. Therefore, for embodiments where the channel semiconductor layer has been designed for a conduction band offset from silicon (e.g., with an InGaN channel) or where the transition layer has a sufficiently large band gap, the silicon fin 210 may not be sacrificial, and the method 101 may then continue to operation 160. However, even where a conduction band offset is present, removing the silicon fin 210 may still improve carrier confinement and / or device performance in other respects (e.g., by enabling a higher breakdown voltage).

[0036] For those in Fig. 3B, the silicon fin 210 is formed in operation 130 ( Fig. 1) etched out of the exposed upper surface to form a gap or trench 330 ( Fig. 3B) between the III-N epitaxial layers. Any of the many chemistries known in the art that etch silicon but not the III-N epitaxial layers may be used in operation 130. Therefore, for embodiments where the channel semiconductor layer has not been designed to be conduction band offset from silicon (e.g., with a GaN channel), the silicon fin 210 is advantageously removed before proceeding to operation 160. According to certain such embodiments, the removal of the sacrificial epitaxial structure occurs after the formation of structure supports along the template fin. In general, such supports may also be formed themselves where subsequent removal (or subsequent conversion) of the silicon fin 210 is optional to fabricate other transistor structures such as the gate stack, gate stack spacers, source / drains, etc. Fig. 2C shows supports 224 in the form of sacrificial structures that may subsequently be used for the formation of transistor source / drain regions or gate stacks at the same location as the sacrificial supports 224. Alternatively, the supports 224 may be permanent structures, where performing silicon fin removal does not pose a significant problem.

[0037] As in Fig. 2C, the supports 224 may extend only along a portion of the length (y-axis) of the fin 210 in order to further process the upper surface of the fin along the lengths 226 before the gate stack and / or source / drain regions are subsequently formed. According to an exemplary embodiment, the three sacrificial supports 224 may be comprised of a sacrificial dielectric and / or polysilicon, which may be deposited in a bulk manner and patterned, for example, using a plasma etch process common in the art. After patterning to form the sacrificial supports 224, dielectric spacers (not shown) may be formed on the sidewalls of the sacrificial supports 224.

[0038] For those in Fig. 3C, the silicon fin 210 is incorporated into a silicon-based dielectric anchor, for example, made of silicon dioxide, silicon nitride, or silicon oxynitride (SiO x N y), for example, by changing the upper fin surface in operation 135 ( Fig. 1) is subjected to a chemical treatment, whereby the dielectric armature 335 ( Fig. 3C) between the III-N epitaxial layers. For such embodiments, the formation of the supports 224 ( Fig. 2C) before or after the conversion of the anchor, or omitted entirely. The smaller silicon volume resulting from the nanoscale dimensioning of the fin facilitates the conversion of the fin into the dielectric anchor. For example, the entire height of the fin, because it is less than 100 nm, can be reduced in process 135 ( Fig. 1) with moderate thermal and / or plasma oxidation and / or nitriding processing conditions. Furthermore, the volume change associated with the oxidation or nitriding of silicon is reduced when the fin width (W F) is at most 20 nm, so that the amount of expansion associated with silicon conversion can be accommodated by strain in the III-N epitaxial layers. Indeed, amorphization of the silicon fin 210 during dielectric conversion may enable a further reduction in the stress (strain) of the III-N epitaxial layers and / or a further reduction in the defect density relative to their "as-grown" state.

[0039] After the patterned epitaxial growth processes have been completed and the silicon template fin has been optionally treated or removed to improve carrier confinement, the method 101 ends with the formation of a conformal gate stack in operation 160. Doping or epitaxy of source-drain regions may then also be carried out on opposite sides of the gate stack, although this is in Fig. 1 is not shown. The formation of the gate stack and / or the source-drain regions can be carried out with the Fig. 2C, which may be done by depositing the gate stack within the lengths 226 or after the carriers 224 have been removed. Fig. 4A, Fig. 4B and Fig. 4C show sectional views of III-N FETs 401, 402, and 403, respectively, along a plane passing through a channel portion of the FET such that the c-axis of the III-N layers aligns with the x-axis in the Fig. 4A - 4C. As in Fig. 4A, is located on each side wall of a silicon fin 210 having a width W F , which is advantageously less than 20 nm, an epitaxial stack with a thickness T epi , which is advantageously less than 200 nm, wherein the thickness T Bof the transition layers 215A, 215B is advantageously smaller than 100 nm. The channel semiconductor layers 217A, 217B each have a 2DEG 219A or 219B, wherein the current flow of the III-N FET 401 is along the y-axis (from the side in Fig. 4A).

[0040] Above the polarization layers 220A, 220B there is a conformal gate dielectric 240, such as one or more layers of silicon nitride (Si x N y), silicon dioxide (SiO2), alumina (Al2O3), Gd2O3, HfO2, high-dielectric-constant silicates such as HfOSiO, TaSiO, AlSiO, and high-dielectric-constant oxynitrides such as HfON, SiON, AlON, ZrSiON, HfSiON, or a Group III-ON, but not limited thereto. According to embodiments, the gate dielectric 240 comprises a dielectric layer to passivate the interface between the gate electrode 250 and c-plane surfaces ({0001} planes) of the III-N epitaxial stack to maintain high channel mobility and reduce gate leakage current. According to one embodiment, the gate dielectric 240 is deposited by ALD for adequate sidewall surface coverage.

[0041] Above the gate dielectric 240 is the gate electrode layer 250, which comprises a work function metal selected to achieve a desired transistor threshold voltage (Vt) (e.g., greater than 0 V, etc.), with exemplary conductive gate materials including tungsten (W), aluminum (Al), titanium (Ti), tantalum (Ta), nickel (Ni), molybdenum (Mo), germanium (Ge), platinum (Pt), gold (Au), ruthenium (Ru), palladium (Pd), iridium (Ir), their alloys, silicides, carbides, nitrides, and phosphides. According to one embodiment, the gate electrode layer 250 is deposited by ALD for adequate sidewall surface coverage.

[0042] For those in the Fig. 4A, Fig. 4B and Fig. In the exemplary embodiments illustrated in Figure 4C, the III-N transistors utilize two identical channel regions (e.g., 2DEGs 219A, 219B) with conduction properties controlled in parallel by the potential at the gate electrode layer 250. Therefore, the effective current carrying width of the 2DEG for the III-N MOS transistors is approximately twice the height of the silicon fin 210 (H F ), on which the III-N epitaxial stacks were grown. Therefore, even with the polar nature of the III-N MOS transistor, multiple transistor channels are formed from each silicon fin.

[0043] As further stated in Fig. 4A, transistor 401 retains silicon fin 210 in the final transistor structure. For the Fig. However, in the transistor 402 shown in Figure 4B, where the silicon fin 210 was removed prior to gate stack formation, the gate dielectric layer 240 directly contacts the junction layers 215A, 215B. The gate electrode layer 250 is further disposed between the junction layers 215A, 215B and is insulated from the epitaxial layers by the conformal gate dielectric layer 240. According to the exemplary embodiment, the isolation dielectric 241 is recessed prior to gate stack formation (e.g., before or after removing the silicon fin), thereby allowing the gate dielectric 240 and the gate metal to fully wrap around the two III-N epitaxial layer stacks. Fig. 4C shows a channel region of a transistor 403 including the silicon-based dielectric fin 335 located between the junction layers 215A and 215B. For such embodiments, the gate dielectric 240 may or may not completely wrap around the pair of III-N epitaxial layer stacks, depending on whether the isolation dielectric 241 is recessed prior to gate stack formation and whether the dielectric fin 335 is undercut during the recess etch.

[0044] Fig. 5 is a functional block diagram of an SoC implementation of a mobile computing platform according to an embodiment of the present invention. The mobile computing platform 500 may be any portable device configured for electronic data display, electronic data processing, or wireless electronic data transmission. For example, the mobile computing platform 500 may be a tablet, a smartphone, a laptop computer, etc., and may include a display screen 505, which, according to the exemplary embodiment, is a touch screen (e.g., capacitive, inductive, resistive, etc.), thereby enabling receipt of user input, the SoC 510, and a battery 513.As illustrated, the greater the level of integration of the SoC 510, the more of the form factor within the mobile computing platform 500 can be occupied by the battery 513 for longest operating times between charges or by storage (not shown) such as an SSD drive for highest functionality.

[0045] Depending on its applications, the mobile computing platform 500 may include other components, including, but not limited to, volatile memory (e.g., DRAM), non-volatile memory (e.g., ROM), flash memory, a graphics processor, a digital signal processor, a cryptoprocessor, a chipset, an antenna, a display, a touchscreen display, a touchscreen controller, a battery, an audio codec, a video codec, a power amplifier, a global positioning system (GPS) device, a compass, an accelerometer, a gyroscope, a speaker, a camera, and a mass storage device (such as a hard disk drive, a compact disk (CD), a digital versatile disk (DVD), etc.).

[0046] The SoC 510 is further shown in the enlarged view 520. Depending on the embodiment, the SoC 510 comprises a portion of a substrate 102 (i.e., a chip) on which two or more of the following are fabricated: a power management integrated circuit (PMIC) 515, an RF integrated circuit (RFIC) 525 having an RF transmitter and / or receiver, a controller 511 thereof, and one or more central processor cores 530, 531. The RFIC 525 may implement any of a number of wireless standards or protocols, including WiFi (IEEE 802.11 family), WiMAX (IEEE 802.16 family), IEEE 802.20, Long Term Evolution (LTE), Ev-DO, HSPA+, HSDPA+, HSUPA+, EDGE, GSM, GPRS, CDMA, TDMA, DECT, Bluetooth, derivatives thereof, as well as any other wireless protocols referred to as 3G, 4G, 5G, etc. The RFIC 525 can contain multiple communication chips.For example, a first communication chip may be provided for shorter-range wireless communications such as WiFi and Bluetooth, and a second communication chip may be provided for longer-range wireless communications such as GPS, EDGE, GPRS, CDMA, WiMAX, LTE, Ev-DO, and others.

[0047] As those skilled in the art will understand, these functionally distinct circuit modules typically use only CMOS transistors, except in the PMIC 515 and the RFIC 525. According to embodiments of the present invention, the PMIC 515 and the RFIC 525 use one or more of the Group III nitride transistors described herein (e.g., Group III nitride transistor 401) using an embodiment of the horizontal c-axis III-N epitaxial stacks described herein. According to further embodiments, the PMIC 515 and the RFIC 525, which use the Group III nitride transistors described herein, are integrated with one or more of the controller 511 and the processor cores 530, 531 provided in silicon CMOS technology monolithically integrated with the PMIC 515 and / or the RFIC 525 on the (silicon) substrate 102.It should be noted that within the PMIC 515 and / or the RFIC 525, the high voltage and high frequency capable Group III nitride transistors described herein need not be used to the exclusion of CMOS, but rather silicon CMOS may be incorporated into either the PMIC 515 or the RFIC 525.

[0048] The Group III nitride transistors described herein can be used particularly where high voltage swings are present (e.g., 7-10 V battery power regulation, DC-to-DC conversion, etc. within the PMIC 515). As explained, according to the exemplary embodiment, the PMIC 515 has an input coupled to the battery 513 and an output that supplies power to all other functional modules in the SoC 510. According to another embodiment, where additional ICs are provided within the mobile computing platform 500 but external to the SoC 510, the output of the PMIC 515 further provides power to all of these ICs external to the SoC 510.

[0049] As further shown, the PMIC according to the exemplary embodiment has an output coupled to an antenna and may further have an input coupled to a communication module on the SoC 510, such as an analog and digital RF baseband module (not shown). Alternatively, these communication modules may be provided on an IC off-chip from the SoC 510 and coupled into the SoC 510 for transmission. Depending on the Group III nitride materials used, the Group III nitride transistors described herein (e.g., III-N transistor 401) may further provide the high power efficiency (PAE) required by a power amplifier transistor with an Ft of at least ten times the carrier frequency (e.g., 1.9 GHz in an RFIC 525 designed for 3G or GSM mobile communications).

[0050] Fig.6 shows a computing device 600 according to one implementation of the invention. A circuit board 602 is housed within the computing device 600. The circuit board 602 may include a number of components, including, but not limited to, a processor 604 and at least one communication chip 606. The processor 604 is physically and electrically coupled to the circuit board 602. According to some implementations, the at least one communication chip 606 is also physically and electrically coupled to the circuit board 602. In other implementations, the communication chip 606 is part of the processor 604.

[0051] Depending on its applications, the computing device 600 may include other components that may be physically and electrically coupled to the board 602.

[0052] These other components include, but are not limited to, volatile memory (e.g., DRAM), non-volatile memory (e.g., ROM), flash memory, a graphics processor, a digital signal processor, a cryptoprocessor, a chipset, an antenna, a display, a touchscreen display, a touchscreen controller, a battery, an audio codec, a video codec, a power amplifier, a global positioning system (GPS) device, a compass, an accelerometer, a gyroscope, a speaker, a camera, and a mass storage device (such as a hard disk drive, a compact disk (CD), a digital versatile disk (DVD), etc.).

[0053] Communication chip 606 enables wireless communications for transferring data to and from computing device 600. The term "wireless" and its derivatives can be used to describe circuits, devices, systems, methods, techniques, communication channels, etc., that can transmit data through the use of modulated electromagnetic radiation over a non-solid medium. The term does not imply that the associated devices do not contain wires, although this could be the case according to some embodiments. The communication chip 606 may implement any of a number of wireless standards or protocols, including WiFi (IEEE 802.11 family), WiMAX (IEEE 802.16 family), IEEE 802.20, Long Term Evolution (LTE), Ev-DO, HSPA+, HSDPA+, HSUPA+, EDGE, GSM, GPRS, CDMA, TDMA, DECT, Bluetooth, derivatives thereof, as well as any other wireless protocols known as 3G, 4G, 5G, etc.Computing device 600 may include, but is not limited to, multiple communication chips 606. For example, a first communication chip 606 may be provided for shorter-range wireless communications, such as Wi-Fi and Bluetooth, and a second communication chip 606 may be provided for longer-range wireless communications, such as GPS, EDGE, GPRS, CDMA, WiMAX, LTE, Ev-DO, and others.

[0054] The processor 604 of the computing device 600 includes an integrated circuit die encapsulated within the processor 604. According to some embodiments of the invention, the processor's integrated circuit die includes one or more MOSFET-type devices constructed according to embodiments described elsewhere herein. The term "processor" may refer to any device or portion of a device that processes electronic data from registers and / or memory to convert that electronic data into other electronic data that may be stored in registers and / or memory.

[0055] The communication chip 606 also includes an integrated circuit die encapsulated within the communication chip 606. According to another embodiment of the invention, the integrated circuit die of the communication chip includes one or more MOSFET-type devices having features according to embodiments described elsewhere herein and / or being fabricated according to embodiments described elsewhere herein.

[0056] According to further implementations, another component incorporated into the computing device 600 may include an integrated circuit die having one or more MOSFET-type devices having features according to embodiments described elsewhere herein and / or being fabricated according to embodiments described elsewhere herein.

[0057] According to embodiments, the computing device 600 may be a laptop, a netbook, a notebook, an ultrabook, a smartphone, a tablet, a personal digital assistant (PDA), an ultra-mobile PC, a mobile phone, a desktop computer, a server, a printer, a scanner, a display, a set-top box, an entertainment control unit, a digital camera, a portable music player, or a digital video recorder.

Claims

[1] III-N field-effect transistor, hereinafter abbreviated III-N FET, arranged on a silicon substrate, the III-N FET comprising: an anchor arranged above the substrate (203), a first and a second Group III-N device layer stack physically separated from each other by the anchor, each stack having a c-axis extending in opposite directions from the anchor and parallel to a surface plane of the substrate (203), and a gate stack disposed over the III-N device layer stacks to control the conductivity of a channel semiconductor layer (217A, 217B) in each of the III-N device layer stacks, the channel semiconductor layer having a conduction band offset from the anchor material. [2] The III-N FET of claim 1, wherein the armature further comprises a silicon fin (210) having sidewalls (210A, 210B) with a (111) or a (110) face, the sidewalls being perpendicular to the c-axis of the III-N device layer stacks. [3] The III-N FET of claim 2, wherein the channel semiconductor layer (217A, 217B) comprises InGaN with at most 20% In. [4] The III-N FET of claim 2, wherein the silicon fin (210) has a minimum top surface dimension of at most 20 nm and wherein the sidewalls (210A, 210B) have a z-height of at most 100 nm. [5] The III-N FET of claim 1, wherein the armature further comprises a fin (335) made of a silicon-based dielectric. [6] The III-N FET of claim 5, wherein the anchor further comprises silicon dioxide in contact with a first crystalline junction layer (215A) disposed between the silicon dioxide and a first (217A) of the semiconductor channel layers, the silicon dioxide being in further contact with a second crystalline junction layer (215B) disposed between the silicon dioxide and a second of the semiconductor channel layers (217B). [7] The III-N FET of claim 1, wherein the armature further comprises a gate dielectric layer (240), and wherein the gate dielectric layer (240) is further disposed over a III-N polarization layer (220A, 220B) of each of the III-N device layer stacks. [8] The III-N FET of claim 7, wherein the armature further comprises a gate electrode material, and wherein the gate dielectric layer (240) is disposed between the gate electrode material and each of the first and second III-N device layer stacks. [9] The III-N FET of claim 1, wherein each of the III-N device layer stacks further comprises: a channel semiconductor layer (217A, 217B) made of GaN arranged on a transition layer (215A, 215B), and a polarization layer (220A, 220B) made of AlN, AlInN, AlGaN or AlInGaN arranged on the GaN channel semiconductor layer. [10] The III-N FET of claim 9, wherein the transition layer (215A, 215B) further comprises at least one of the following: a crystalline oxide, AlN, AlInN, or AlGaN. [11] A mobile computing device (600) comprising: a touch screen, a battery, an antenna, a DC-DC converter coupled to the battery, and a wireless transmitter further comprising a power amplifier, PA, wherein at least one of the DC-DC converter and the PA comprises the III-N FET of claim 1. [12] The mobile computing device (600) of claim 11, wherein the DC-DC converter comprises a first III-N FET of claim 1 and the PA uses a second III-N FET of claim 1. [13] A method (101) for forming a III-N field effect transistor (FET), the method comprising the following steps: Forming (110) a silicon fin over a substrate, the silicon fin having first and second sidewalls with a (111) surface or a (110) surface, epitaxially growing (115) a first crystalline transition layer on the first sidewall and a second crystalline transition layer on the second sidewall, epitaxially growing (118, 117) a first III-N semiconductor channel layer over the first transition layer and a second III-N semiconductor channel layer over the second transition layer, epitaxially growing (121, 120) a first III-N semiconductor polarization layer over the first III-N semiconductor channel and a second III-N semiconductor polarization layer over the second III-N semiconductor channel, and Forming (160) a gate stack over the first and second polarization layers, the method comprising at least one of the following steps: Growing (117) an InGaN material as the III-N semiconductor channel layers, Removing (130) the silicon fin following the epitaxial growth of at least the transition layers or Converting (135) the silicon fin into a silicon-based dielectric material subsequent to epitaxially growing at least the transition layer. [14] The method of claim 13, wherein the method comprises: removing the silicon fin with a selective etch chemistry after epitaxially growing the junction layers, the channel layers, and the polarization layers. [15] The method of claim 13, wherein the method comprises: converting the silicon fin into a silicon-based dielectric after epitaxially growing the junction layers, the channel layers, and the polarization layers. [16] The method of claim 15, wherein converting the silicon fin into a silicon-based dielectric further comprises subjecting an upper surface of the silicon fin to a plasma or thermal oxidation process. [17] The method of claim 13, wherein forming the fin further comprises: Structuring an isolation region around a portion of the substrate silicon with a width of at most 20 nm and Removing the isolation region to provide the first and second sidewalls having a height of at most 100 nm, and wherein the method further comprises: forming a structural support along a length of the fin prior to removing the silicon fin or converting the silicon fin into the silicon-based dielectric material. [18] The method of claim 13, wherein forming the gate stack further comprises: Depositing one or more gate dielectric layers over the polarization layers using an atomic layer deposition process and Depositing one or more gate electrode layers over the gate dielectric layer using an atomic layer deposition process. [19] The method of claim 18, wherein depositing the one or more gate dielectric layers further comprises: depositing the one gate dielectric layer on a side of the junction layer opposite the III-N semiconductor channel layer. [20] The method of claim 13, wherein epitaxially growing the first and second crystalline transition layers further comprises growing at least one of a crystalline oxide and a III-N semiconductor to a thickness of less than 100 nm, wherein the first and second semiconductor channel layers are grown to a thickness of less than 50 nm, and wherein the first and second polarization layers are grown to a thickness of less than 10 nm.

Citation Information

Patent Citations

  • Method for manufacturing FinFET structures with strain-inducing source / drain-forming spacers and FinFET structures

    DE112010002352B4

  • Semiconductor device and method of manufacturing semiconductor device

    US20070004117A1

  • INTEGRATED METHOD FOR FORMING METAL GATE FinFET DEVICES

    US20120015493A1

  • Epitaxy Profile Engineering for FinFETs

    US20130001705A1