Silicon carbide semiconductor device and fabrication method thereof

Abstract

A silicon carbide semiconductor device has a first-conductivity-type semiconductor layer having a lower impurity concentration and formed on a first-conductivity-type semiconductor substrate, a second-conductivity-type semiconductor layer having a higher impurity concentration and selectively formed in the first-conductivity-type semiconductor layer, a second-conductivity-type base layer having a lower impurity concentration formed on a surface of the second-conductivity-type semiconductor layer, a first-conductivity-type source region selectively formed in a surface layer of the base layer, a first-conductivity-type well region formed to penetrate the base layer from a surface to the first-conductivity-type semiconductor layer, and a gate electrode formed via a gate insulation film on a surface of the base layer interposed between the source region and the well region. Portions of the respective second-conductivity-type semiconductor layers of different cells can be connected to each other by a connecting portion in a region under the well region.

Claims

1. A silicon carbide semiconductor device having a semiconductor device structure created inside a semiconductor substrate, the silicon carbide semiconductor device comprising: a gate pad configured to establish electric contact with an outside electrode, wherein the semiconductor device structure is formed in the semiconductor substrate at a lower portion beneath the gate pad, the semiconductor device structure includes: a first-conductivity-type semiconductor substrate, a first-conductivity-type semiconductor layer having a first impurity concentration and formed on the semiconductor substrate, a second-conductivity-type semiconductor layer having a second impurity concentration and selectively formed in the first-conductivity-type semiconductor layer, a second-conductivity-type base layer having a third impurity concentration and formed on a surface of the second-conductivity-type semiconductor layer, wherein the second impurity concentration is a greater concentration than the first impurity concentration and the third impurity concentration, a first-conductivity-type source region selectively formed in a surface layer of the base layer, a first-conductivity-type well region formed to penetrate from the surface layer of the base layer to the first-conductivity-type semiconductor layer, and a gate electrode formed over a gate insulation film on a surface of the base layer interposed between the source region and the well region, a plurality of the semiconductor device structure constituting a plurality of different cells, portions of the respective second-conductivity-type semiconductor layers of the different cells are connected to each other in a region under the respective well regions, wherein a portion of the source region under the gate pad is electrically connected to a portion of the source region of the semiconductor device structure other than the portion of the source region under the gate pad, the source region is surrounded by a second-conductivity-type semiconductor region, and a first area ratio of the portion of the source region under the gate pad compared with a whole portion under the gate pad is larger than a second area ratio of a portion of the source region under a source pad compared with a whole portion under the source pad.

2. The silicon carbide semiconductor device according to claim 1, wherein a crystallographic plane index of the first-conductivity-type semiconductor substrate indicates a plane parallel to, or a plane tilted within 10 degrees relative to, a (000-1) plane.

3. The silicon carbide semiconductor device according to claim 1, wherein a crystallographic plane index of the first-conductivity-type semiconductor substrate indicates a plane parallel to, or a plane tilted within 10 degrees relative to, a (0001) plane.

4. The silicon carbide semiconductor device according to claim 1, wherein the source region of the gate pad portion is formed into a linear shape.

5. The silicon carbide semiconductor device according to claim 1, wherein the source region of the gate pad portion is formed into a polygonal mesh shape.

Description

BRIEF DESCRIPTION OF DRAWINGS

(1) FIG. 1 is a cross-sectional view at fabrication steps of an SiC-MOSFET of a first embodiment of the present invention;

(2) FIG. 2 is a plane view of arrangement of the P.sup.+ layers and cells of the SiC-MOSFET of the first embodiment of the present invention;

(3) FIG. 3 is a chart of measurement results of electric characteristics of SiC-MOSFETs formed according to embodiments of the present invention;

(4) FIG. 4 is a chart of actual measurement results of element breakdown voltage when the element breakdown voltage and the n-inversion layer width are changed in the first embodiment and a comparison example;

(5) FIG. 5 is a chart of measurement results of load short circuit capability of the SiC-MOSFET of the first embodiment;

(6) FIG. 6 is a chart of evaluation results of turn-off failure capability of the SiC-MOSFET of the first embodiment;

(7) FIGS. 7A, 7B, and 7C are diagrams of a device structure under the gate pad of the SiC-MOSFET of the first embodiment;

(8) FIGS. 8A and 8B are diagrams of another device structure under the gate pad of the SiC-MOSFET of the first embodiment;

(9) FIG. 9 is a diagram of arrangement of the P.sup.+ layers and cells of the SiC-MOSFET of the third embodiment of the present invention;

(10) FIGS. 10A and 10B are charts of turn-on switching waveforms of the SiC-MOSFETs of the embodiments;

(11) FIGS. 11A and 11B are charts of turn-off switching waveforms of the SiC-MOSFETs of the embodiments;

(12) FIG. 12 is a cross-sectional view of a general MOSFET;

(13) FIG. 13 depicts a cross-sectional structure of a typical element;

(14) FIG. 14 depicts a cross-sectional structure of a typical element (multi-epi method); and

(15) FIG. 15 depicts a cross-sectional structure of a typical element (trench embedding method).

BEST MODE(S) FOR CARRYING OUT THE INVENTION

(16) Embodiments of a silicon carbide semiconductor device and a fabrication method thereof according to the present invention will be described in detail with reference to the accompanying drawings.

First Embodiment

(17) FIG. 1 is a cross-sectional view at fabrication steps of an SiC-MOSFET of a first embodiment of the present invention. In FIG. 1, (a) is a cross-sectional view of a portion in which P.sup.+ layers are not connected, while (b) is a cross-sectional view of a portion in which the P.sup.+ layers are connected.

(18) In this embodiment, a vertical planar gate MOSFET is described as a MOSFET with an element breakdown voltage of 1200 V formed by using silicon carbide (SiC) as a semiconductor material. First, as depicted in (1), an n.sup.+ type (first-conductivity-type) SiC semiconductor substrate 1 is prepared. In the case of an IGBT, a p.sup.+ type semiconductor substrate 1 is used. In this example, the low-resistance SiC semiconductor 1 is used that includes about 210.sup.19 cm.sup.3 of nitrogen as impurities. An n.sup. type (first-conductivity-type) SiC layer 2 including about 1.810.sup.16 cm.sup.3 of nitrogen is laminated by epitaxial growth to about 10 m on a plane of the n.sup.+ type semiconductor substrate 1 having a crystallographic plane index tilted by 4 degrees relative to a (000-1) plane.

(19) As depicted in (2), a second-conductivity-type P.sup.+ layer 3 is formed with a width of 13 m and a depth of 0.5 m by the ion implantation method on the SiC layer 2. For example, aluminum ions are used in this case. A dosage is set so as to achieve an impurity concentration of 1.010.sup.18 cm.sup.3. Portions of the P.sup.+ layers 3 are connected between different cells 20 under a first-conductivity-type n-inversion layer 6 described later (see (4) of FIG. 1(b) and see FIG. 2 for a plane view). A distance between unconnected positions of the P.sup.+ layers 3 is set to 2 m.

(20) As depicted in (3), a second-conductivity-type P-base layer 4 is subsequently formed with a thickness of 0.5 m on the P.sup.+ layer 3 and the n.sup. type SiC layer 2 by the epitaxial growth method. The impurities in this case are aluminum and an impurity concentration is set to 2.010.sup.16 cm.sup.3.

(21) As depicted in (4), nitrogen ions are selectively implanted at 5.010.sup.16 cm.sup.3 to 1.5 m in depth and 2.0 m in width as the n-inversion layer 6. As depicted in (5), a first-conductivity-type n.sup.+ source layer 7 and a second-conductivity-type p.sup.+ contact layer 8 are selectively formed in the p-base layer 4.

(22) Activation annealing is then performed. Heat treatment temperature and time are 1620 degrees C. and two minutes, respectively. As a result, as depicted in (6), a gate oxidation film is formed with a thickness of 100 nm by thermal oxidation and is annealed near 1000 degrees C. in a hydrogen atmosphere. A polycrystalline silicon layer doped with phosphorus is formed as a gate electrode 9 and patterned on the gate oxidation film.

(23) Subsequently, as depicted in (7), a 1.0-m-thick film of phosphorus glass is formed as an interlayer insulation film 10, which is patterned and then heat-treated. A 5-m-thick film of aluminum including 1% silicon is formed on the surface by the sputtering method to form a surface electrode (source electrode) 11. A film of nickel is formed on the back surface of the element and, after heat treatment at 970 degrees C., a back surface electrode 12 made of titanium, nickel, and gold is formed. A protection film is then added to the surface to complete the element.

(24) FIG. 2 is a plane view of arrangement of the P.sup.+ layers and cells of the SiC-MOSFET of the first embodiment of the present invention. In the example of FIG. 2, the P.sup.+ layers 3 are connected by a connecting portion 3a from each corner portion toward a corner portion of another one of the cells 20 under the n-inversion layer 6. In the example of FIG. 2, the cells 20 are formed in a hexagonal cell pattern. This is not a limitation and the cells 20 may be square cells.

(25) FIG. 3 is a chart of measurement results of electric characteristics of SiC-MOSFETs formed according to embodiments of the present invention. The die size of the embodiments of the present invention is 3 mm square with an active area of 5.27 mm.sup.2 and rated current of 25 A. The ON-resistance (RonA) is a sufficiently low value of 2.8 mcm.sup.2 and the initial element breakdown voltage is 1450 V, which is sufficiently good characteristic for a 1200-V element.

(26) Measurement of a comparison example (conventional technique) was performed by using an SiC-MOSFET created without connection between the P.sup.+ layers 3. Although ON-resistance was a comparable sufficiently low value of 2.8 mcm.sup.2 in the case of the comparison example, the gate oxide film was destroyed when 880 V was applied between the source and the drain. This reveals that the present invention exhibits extremely low ON-resistance while maintaining sufficient element breakdown voltage.

(27) FIG. 4 is a chart of actual measurement results of element breakdown voltage when the element breakdown voltage and the n-inversion layer width are changed in the first embodiment and a comparison example. The comparison example is the SiC-MOSFET without connection of the P.sup.+ layer 3 between cells as described above. Concentration and thickness of each layer of the elements are as described above. As a result, it can be seen that the first embodiment of the present invention realizes a high breakdown voltage characteristic of 1400 V or higher, which is a sufficient breakdown voltage characteristic for a 1200-V device. The ON-resistance is the same in both the first embodiment and the comparison example and it is found that creating a cell under a gate pad as in the first embodiment is effective for reducing the ON-resistance.

(28) It is found that, to satisfy the high breakdown voltage characteristic of 1400 V or higher equivalent to the first embodiment in the SiC-MOSFET of the comparison example, a distance between the P.sup.+ layers 3 must be equal to or less than 1.0 m while the inversion concentration of the n-inversion layer 6 must be reduced to one fifth. The ON-resistance in the comparison example of this condition indicates an extremely high value of 10.8 mcm.sup.2. Therefore, the present invention has the ON-resistance smaller than the comparison example and can improve the element breakdown voltage characteristic at the same time.

(29) FIG. 5 is a chart of measurement results of load short circuit capability of the SiC-MOSFET of the first embodiment. In a load short circuit capability test, source voltage is directly applied between the source and the drain and a voltage Vg=20 V is applied to the gate electrode in this state to evaluate a period without destruction of the gate oxidation film in sec. Conditions during the measurement were a source voltage Vcc=800 V and a measurement temperature (Tj) of 175 degrees C. As indicated by measurement waveforms of FIG. 5, the element was not destroyed even when the element conducted the maximum current of 125 A, which is five times larger than the element rating, and was not destroyed even after 15 sec, exhibiting sufficient characteristics.

(30) FIG. 6 is a chart of evaluation results of turn-off failure capability of the SiC-MOSFET of the first embodiment. When the turn-off capability was evaluated, a source-drain voltage was clamped at 1630 V (Vds clamp of FIG. 6) and it was confirmed that the SiC-MOSFET was able to be turned off at 100 A (four times larger than rated current) at 150 degrees C. without destruction. Therefore, it is understood that the element of the first embodiment is an element realizing low ON-resistance and having extremely large load short circuit capability and turn-off capability. When the capabilities of the SiC-MOSFET of the comparison example were evaluated, the results of both the load short circuit capability and the turn-off capability were quite inferior to the element of the first embodiment because of insufficient element breakdown voltage (see, FIG. 3).

(31) Even when elements were created by forming the SiC layer 2 in the same way on the planes of the n-type semiconductor substrate 1 having the crystallographic plane index tilted by 0, 2, 8, and 10 degrees relative to the (000-1) plane, the characteristics were almost unchanged and were favorable.

(32) FIGS. 7A, 7B, and 7C are diagrams of a device structure under the gate pad of the SiC-MOSFET of the first embodiment. FIG. 7A is a plane view; FIG. 7B is a cross-sectional view taken along a cut line A-A of FIG. 7A; FIG. 7C is a cross-sectional view taken along a cut line B-B of FIG. 7A; and the gate electrode and the oxide film are not depicted. To further reduce the ON-resistance in the configuration of the first embodiment, as depicted in FIGS. 7A, 7B, and 7C, a source region 70 is formed under the gate pad such that the source region 70 is electrically connected from a source region 71 under a source pad by using an n-type semiconductor. This n-type semiconductor source region 70 must be separated by a p-type semiconductor region 72 from an n-type semiconductor region 73 acting as a well. In FIGS. 7A, 7B, and 7C, the source region 70 under the gate pad has a linear shape and a p-layer (second-conductivity-type region) 74 for protecting a gate oxide film is disposed orthogonally to the source region 70.

(33) FIGS. 8A and 8B are diagrams of another device structure under the gate pad of the SiC-MOSFET of the first embodiment. FIG. 8A is a plane view; FIG. 8B is a cross-sectional view taken along a cut line C-C of FIG. 8A; and the gate electrode and the oxide film are not depicted. The source region 71 under the gate pad is formed into a hexagonal mesh shape. The p-layer 74 for protecting the gate oxide film is floated at the center of the hexagonal source region 70. The shape of the source region 71 is not limited to a hexagonal shape and may be a polygonal shape.

(34) As depicted in FIGS. 7A, 7B, 7C, 8A and 8B, by fabricating an element under the gate pad, an element utilization area can be increased within the area of the same device and the ON-resistance can be reduced. As seen therein, a first area ratio of the portion of the source region under the gate pad compared with a whole portion under the gate pad is larger than a second area ratio of a portion of the source region under a source pad compared with a whole portion under the source pad.

Second Embodiment

(35) In a second embodiment of the present invention, a 1200-V 25-A SiC-MOSFET was formed by the same fabrication steps as the first embodiment. In the second embodiment, the n-type SiC layer 2 including about 1.810.sup.16 cm.sup.3 of nitrogen was epitaxially grown to about 10 m on the plane of the n-type semiconductor substrate 1 having the crystallographic plane index tilted by 4 degrees relative to the (0001) plane. The other steps and the cell structure were completely the same as the first embodiment. As described in the electric characteristic evaluation results of the element of the second embodiment depicted in FIG. 3, although the ON-resistance is increased by about 55% as compared to the first embodiment, it can be seen that the element has the sufficiently low ON-resistance characteristic as compared to normal SiC-MOSFETs. When the n-type SiC layer 2 was formed in the same way on the planes of the n-type semiconductor substrate 1 having the crystallographic plane index tilted by 0, 2, 8, and 10 degrees relative to the (0001) plane and the element evaluation was performed for the created elements, the characteristics were almost unchanged and were favorable.

Third Embodiment

(36) In a third embodiment of the present invention, a 1200-V 25-A SiC-MOSFET was formed by the same fabrication steps as the first embodiment. The n-type SiC layer 2 including about 1.810.sup.16 cm.sup.3 of nitrogen was epitaxially grown to about 10 m on the plane of the n-type semiconductor substrate 1 having the crystallographic plane index tilted by 4 degrees relative to the (000-1) plane.

(37) FIG. 9 is a diagram of arrangement of the P.sup.+ layers and cells of the SiC-MOSFET of the third embodiment of the present invention. As depicted in FIG. 9, the cells are formed in a stripe pattern. Therefore, the arrangement of the P.sup.+ layers 3 allows the P.sup.+ layers 3 to be connected by the connecting portion 3a between the cells 20, 20. The other steps are the same as the first embodiment. As described in the electric characteristic evaluation results of the formed element depicted in FIG. 3, although the ON-resistance is increased by about 10% as compared to the first embodiment, it can be seen that the element has the sufficiently low ON-resistance characteristic and the high breakdown voltage characteristic as compared to normal SiC-MOSFETs.

Fourth Embodiment

(38) A fabrication method a fourth embodiment of the present invention will be described. First, the n-type SiC semiconductor substrate 1 is prepared. The low-resistance SiC semiconductor 1 is used that includes about 210.sup.19 cm.sup.3 of nitrogen as impurities. The n-type SiC layer 2 including about 1.810.sup.16 cm.sup.3 of nitrogen is epitaxially grown to about 10 m on a plane of the n-type semiconductor substrate 1 having the crystallographic plane index tilted by 4 degrees relative to a (000-1) plane. The P.sup.+ layer 3 is formed with a width of 13 m and a depth of 0.5 m by the epitaxial method on the n-type SiC layer 2. Aluminum was used for impurity ions in this case. A dosage is set so as to achieve an impurity concentration of 1.010.sup.18 cm.sup.3. As is the case with the first embodiment, portions of the P.sup.+ layers 3 are connected to each other under the n-inversion layer 6 (see FIG. 2). Although the cells were formed in a hexagonal cell pattern in the fourth embodiment, the cells may be square cells etc. A distance between the P.sup.+ layers 3 at unconnected positions of the P.sup.+ layers 3 is set to 2 m.

(39) The P-base layer 4 is subsequently formed with a thickness of 0.5 m on the P.sup.+ layer 3 and the n-type SiC layer 2 by the epitaxial growth method. The impurities in this case are aluminum and an impurity concentration is set to 2.010.sup.16 cm.sup.3. Nitrogen ions are selectively implanted as the n-inversion layer 6, and the n.sup.+ source layer 7 and the p.sup.+ contact layer 8 are selectively formed in the p-base layer 4. The n-inversion layer 6 has the concentration, thickness, and width same as the first embodiment.

(40) Activation annealing is then performed. Heat treatment temperature and time are 1620 degrees C. and two minutes, respectively. The gate oxidation film is formed with a thickness of 100 nm by thermal oxidation and is annealed near 1000 degrees C. in a hydrogen atmosphere. After a polycrystalline silicon layer doped with phosphorus is formed and patterned as a gate electrode, a 1.0-m-thick film of phosphorus glass is formed as the interlayer insulation film 10 and is patterned and heat-treated, and a 5-m-thick film of aluminum including 1% silicon is formed on the surface by the sputtering method. A film of nickel is formed on the back surface of the element and, after heat treatment at 970 degrees C., the back surface electrode 12 made of titanium, nickel, and gold is formed. A protection film is then added to the surface to complete the element.

(41) FIG. 3 depicts measurement results of electric characteristics of the SiC-MOSFET of the fourth embodiment formed as described above. A die size is 3 mm square with an active area of 5.27 mm.sup.2 and rated current of 25 A. The ON-resistance (RonA) is a sufficiently low value of 2.85 mcm.sup.2 and the initial element breakdown voltage is 1455 V, which is sufficiently good characteristic for a 1200-V element. When films were formed in the same way on the planes of the n-type semiconductor substrate 1 having the crystallographic plane index tilted by 0, 2, 8, and 10 degrees relative to the (000-1) plane and the element evaluation was performed for the created elements, the characteristics were almost unchanged and were favorable.

Fifth Embodiment

(42) In a fifth embodiment, a 1200-V 25-A SiC-MOSFET was produced by the same fabrication steps as the fourth embodiment. However, in the fifth embodiment, the n-type SiC layer 2 including about 1.810.sup.16 cm.sup.3 of nitrogen was epitaxially grown to about 10 m on the plane of the n-type semiconductor substrate 1 having the crystallographic plane index tilted by 4 degrees relative to the (0001) plane. The other steps were completely the same as the fourth embodiment. The electric characteristic evaluation results of the produced element are depicted in FIG. 4. Although the ON-resistance is increased by about 50% as compared to the fourth embodiment, it can be seen that the element has the sufficiently low ON-resistance characteristic as compared to normal SiC-MOSFETs. When films were formed on the planes of the n-type semiconductor substrate 1 having the crystallographic plane index tilted by 0, 2, 8, and 10 degrees relative to the (0001) plane and the element evaluation was performed for the formed elements, the characteristics were almost unchanged and were favorable.

(43) FIGS. 10A and 10B are charts of turn-on switching waveforms of the SiC-MOSFETs of the embodiments and FIGS. 11A and 11B are charts of turn-off switching waveforms of the SiC-MOSFETs of the embodiments. The charts represent the switching loss evaluations of the SiC-MOSFETs formed in the first to fifth embodiments. The measurement results of FIGS. 10A and 11A were acquired at room temperature and the measurement results of FIGS. 10B and 11B were acquired at a temperature of 200 degrees, respectively. The turn-on loss and the turn-off loss were able to be reduced and the favorable waveforms were acquired from both the turn-on switching and the turn-off. According to the SiC-MOSFET of the present invention, as depicted in FIG. 3, both the turn-on and turn-off losses can be reduced by 60% or more as compared to an Si-IGBT of the same rating (1200 V and 25 A).

(44) According to the configuration, if the impurity concentrations are significantly increased in the n-type semiconductor layer and the n-type semiconductor well region to sufficiently reduce the ON-resistance, or if the distances are made larger between the p-type semiconductor layers and between the base layers to sufficiently reduce the ON-resistance, or if a higher voltage is applied between the source and the drain (the source is 0 V and a positive voltage is applied to the drain), a large electric field is not applied to the gate oxide film on the n-type semiconductor well region and a sufficient element breakdown voltage can be retained in any case. This is because a depletion layer can easily spread in the lateral direction along the P.sup.+ contact layer. As a result, because of the design facilitating the spread of the depletion layer even when the impurity concentrations of the n-type semiconductor layer and the n-type semiconductor well region are set to be higher than conventional Si-MOSFETs, the distances can be made larger between the p-type contact layers and between the p-type base layers to make the ON-resistance smaller while the element breakdown voltage is sufficiently maintained.

(45) If the p-type base layer is formed by the epitaxial growth method, the base layer can be made flat almost without surface roughness and therefore, the mobility of a MOSFET portion on the surface becomes extremely large and, as a result, the ON-resistance can be made further smaller.

(46) If silicon carbide is used as the semiconductor material, a crystallographic plane index of the n-type semiconductor substrate can be set to a plane parallel to, or within 10 degrees relative to, the (000-1) plane, or a crystallographic plane index of the n-type semiconductor substrate can be set to a plane parallel to, or within 10 degrees relative to, the (0001) plane to reduce the interface state density at the interface between the gate oxide film and the semiconductor and to further improve the mobility of the MOSFET portion. As a result, the ON-resistance can be made extremely small.

(47) By achieving a structure in which the source region is formed to create an element structure also under the gate pad, the effective element area can be increased to make the ON-resistance smaller without the need for a complicated layered electrode structure.

(48) Although the silicon carbide semiconductor device of the present invention has been described by taking a MOSFET as an example in the embodiments, this is not a limitation. For example, the present invention is also applicable to an IGBT and the IGBT can have low ON-resistance with reliability without destruction of a gate oxide film even when a high voltage is applied.

(49) The present invention enables the provision of power devices such as MOSFETs and IGBTs having low ON-resistance and large critical electric field and capable of achieving high-speed switching characteristics while retaining sufficient element breakdown voltage characteristics regardless of crystal plane orientation of the substrate.

INDUSTRIAL APPLICABILITY

(50) As described above, the present invention is generally applicable to power devices using SiC substrates and is useful for manufacturing of MOSFETs and IGBTs.

EXPLANATIONS OF LETTERS OR NUMERALS

(51) 1 SiC semiconductor substrate 2 SiC layer 3 P.sup.+ layer 4 base layer 6 n-inversion layer 7 source layer 8 contact layer 11 source electrode 12 back surface electrode 20 cell