阅读说明：本技术 具有低栅极电荷和低品质因数的功率器件 (Power device with low gate charge and low quality factor ) 是由 高阳 于 2020-03-09 设计创作，主要内容包括：本公开涉及具有低栅极电荷和低品质因数的功率器件。器件包括单元,其中每个单元包括：本体,该本体具有主顶表面和主底表面；栅极,该栅极在器件的主表面上,并且具有第一长度；栅极绝缘层,该栅极绝缘层在栅极之上,并且具有第二长度,第二长度是第一长度的至少两倍；源极接触部,该源极接触部在器件本体中邻近于栅极；源极金属层,该源极金属层在栅极绝缘层之上；以及漏极,该漏极在单元的主底表面上。(The present disclosure relates to power devices with low gate charge and low quality factor. The device includes cells, wherein each cell includes: a body having a major top surface and a major bottom surface; a gate on a major surface of the device and having a first length; a gate insulating layer over the gate electrode and having a second length that is at least twice the first length; a source contact in the device body adjacent to the gate; a source metal layer over the gate insulating layer; and a drain on the major bottom surface of the cell.)

1. A device comprising cells, wherein each cell comprises:

a body comprising a major top surface and a major bottom surface;

a gate on the major surface of the device and having a first length;

a gate insulating layer over the gate and having a second length that is at least twice the first length;

a source contact on a side of the device body adjacent to the gate;

a source metal layer over the gate insulating layer; and

a drain on the major bottom surface of the cell.

2. The device of claim 1, wherein the device comprises a plurality of substantially identical cells.

3. The device of claim 1, wherein the gate is asymmetric with respect to the gate insulation layer.

4. The device of claim 1, wherein the source contact is asymmetric with respect to the gate insulation layer.

5. The device of claim 1, wherein the second length is at least three times the first length.

6. The device of claim 1, wherein the gate comprises a minimum length gate.

7. The device of claim 1, wherein the gate comprises a polysilicon gate.

8. A device comprising cells, wherein each cell comprises:

a gate strip having a first edge and a second edge;

a source strip on a side of the device cell, the source strip extending along and overlapped by the first edge of the gate strip; and

a charge reduction strip extending along the second edge of the gate strip.

9. The device of claim 8, wherein the device comprises a plurality of substantially identical cells.

10. The device of claim 8, further comprising: a channel under the gate strip in an active mode of operation.

11. The device of claim 8, wherein no channel is formed under the charge reduction strip in an active mode of operation.

12. The device of claim 8, wherein a length of the charge reduction strip is greater than a length of the source strip or the gate strip.

13. The device of claim 8, wherein the gate strips comprise minimum length gate strips.

14. The device of claim 8, wherein the gate strips comprise polysilicon gate strips.

15. A method of fabricating a device cell, the method comprising:

forming a gate having a first length on a major surface of the device cell;

forming a gate insulating layer over the gate, the gate insulating layer having a second length that is at least twice the first length;

forming a source contact on a side of the device cell adjacent to the gate;

forming a source metal layer over the gate insulating layer; and

a drain is formed on a major bottom surface of the device cell.

16. The method of claim 15, further comprising forming a plurality of substantially identical cells.

17. The method of claim 15, wherein the gate is formed asymmetrically with respect to the gate insulation layer.

18. The method of claim 15, wherein the source contact is formed asymmetrically with respect to the gate insulation layer.

19. The method of claim 15, wherein forming the gate comprises forming a minimum length gate.

20. The method of claim 15, wherein forming the gate comprises forming a polysilicon gate.

Technical Field

The present invention generally relates to a power device including a low figure of merit ("FOM").

Background

For VDMOS (vertical, double-diffused, metal oxide, semiconductor) devices used as power switches, the gate charge Qg plays an important role in generating switching losses. Generally, for maximum switching performance, it is desirable to obtain the lowest gate charge Qg. Therefore, RQg FOM determines switching performance in terms of conduction and gate drive power loss, and is represented by the equation RQg FOM ═ (rds (on) × Qg). As noted previously, the FOM should be minimized. Current VDMOS have a relatively high Qg and therefore a high FOM, which can lead to power losses, especially for applications where switching losses dominate.

One method of evaluating MOSFET (metal oxide semiconductor field effect transistor) devices is to compare FOMs or figures of merit. A simple FOM is Rfg FOM. In its simplest form, the RQg figure of merit includes the gate charge (Qg) multiplied by the "on" resistance (rds (on)) between the drain and source of the device. The result of this multiplication generates a RQg FOM, which can then be used in a compare device or some device technology. Generally, a lower RQg FOM corresponds to lower switching losses.

Several approaches have been attempted to reduce Qg without significantly increasing rds (on). One typical approach is to have a planar "split gate" structure by removing the gate polysilicon over the JFET (junction field effect transistor) neck (tack). In this case, the gate-drain overlap will be significantly reduced. Although both Qg (gate charge) and Qgd (gate drain charge) will decrease, rds (on) increases. In addition, the "split gate" approach has minimal impact on Qgs (gate-source charge), which is also important in determining device switching performance.

Disclosure of Invention

The device includes cells, wherein each cell includes: a body comprising a major top surface and a major bottom surface; a gate on a major surface of the device and having a first length; a gate insulating layer over the gate electrode and having a second length that is at least twice the first length; a source contact in the device body adjacent to the gate; a source metal layer over the gate insulating layer; and a drain on the major bottom surface of the cell.

Drawings

For a more complete understanding of the present invention and the advantages thereof, reference is now made to the following descriptions taken in conjunction with the accompanying drawings, in which:

fig. 1A is a cross-sectional view of a super junction ("SJ") VDMOS device according to the prior art;

FIG. 1B is a cross-sectional view of the SJ VDMOS device of FIG. 1A, including an indication of vertical current flow within the device in an active mode of operation;

FIG. 2A is a cross-sectional view of an SJ VDMOS device including a modified FOM according to one embodiment;

fig. 2B is a cross-sectional view of the SJ VDMOS device of fig. 2A, including an indication of vertical current within the device in an active mode of operation;

FIG. 3 is a plan view of the device of FIGS. 1A and 1B;

FIG. 4 is a plan view of the device of FIGS. 2A and 2B arranged in a first configuration, according to one embodiment;

FIG. 5 is a plan view of the device of FIGS. 2A and 2B arranged in a second configuration, according to one embodiment;

fig. 6 is a table of simulation results for a plurality of 100V SJ VDMOS devices including different polysilicon gate lengths, in accordance with an embodiment; and is

Fig. 7 is a table of simulation and measurement results for a plurality of 250V SJ VDMOS devices including different polysilicon gate lengths, according to an embodiment.

Detailed Description

According to embodiments, methods, transistor cells, and transistor devices including a plurality of transistor cells are described that are configured to reduce Qg without significantly increasing rds (on), thereby improving FOM. By removing one of the two source contacts in a conventional SJ VDMOS device cell while maintaining the other device structure, the polysilicon gate length is significantly reduced and the FOM is improved. The embodiment concepts may also be adapted to other device types. For example, any vertical device having two symmetric source contacts may be modified as described herein. According to embodiments, a device cell may be replicated in two or more configurations while maintaining the improvement in FOM.

Fig. 1A is a cross-sectional view of an exemplary SJ VDMOS device cell 100 that includes a source metal region 102, a gate insulation layer 104, an insulated gate 106, source contacts 108A and 108B, a body region including regions 110 and 114, an interface 112 between regions 110 and 114, and a drain contact 116. The source metal region 102 may be formed of aluminum or other conductive material. A dielectric layer such as silicon dioxide or other dielectric material may be used for the gate insulation layer 104. The insulated gate 106 may comprise a polysilicon or aluminum gate surrounded by a dielectric layer such as silicon dioxide or other dielectric material. The source contacts 108A and 108B may include diffusion regions, such as N + type diffusion regions, in the body of the device cell 100. The source contacts 108A and 108B may be diffused into another diffusion region, such as a P-base region for forming a channel region. Region 110 may include P-type pillars and region 114 may include an N-type epitaxial region. Regions 110 and 114 will support a reverse bias voltage when the device is in the off state. Interface 112 includes a PN junction between regions 110 and 114, where depletion regions are formed on each side of junction 112 when the device is in an off state. The drain contact 116 may be a metalized bottom surface of the device cell 100, such as aluminum or an aluminum alloy. The body of device cell 100 may include an N-drift region in an epitaxial region over an N + substrate. Region 110 also includes a P + contact 103, which is shorted to source metal region 102 as shown.

Fig. 1B is a cross-sectional view of the SJ VDMOS device cell 100 of fig. 1A, including indications of vertical currents 118, 120, and 122 within the device in an active mode of operation. In order to simplify the drawing understanding, not all reference numerals are repeated in fig. 1B. The main path 122 of current through the body of the device cell 100 splits into two equal branches, including current branch 118 and current branch 120, current branch 118 flowing up the main top surface of the device and through the first channel to the first source contact 108A, and current branch 120 flowing up the main top surface of the device and through the second channel to the second source contact 108B. Current branch 122 represents the total vertical current through the device, and current branches 118 and 120 each represent approximately half of the total vertical current through the device.

Thus, the device cell 100 includes N + source contacts 108A and 108B on both sides of the polysilicon gate 106. During conduction, a drain current flows to each of the N + source contacts through the left and right channels.

A plurality of device cells are replicated in a pattern, such as rows and columns, and interconnected to complete the entire power device, as is known in the art.

Fig. 2A is a cross-sectional view of a SJ VDMOS device cell 200 including a modified FOM, including a source metal region 202, a gate insulating layer 204, a single insulated gate 206, a single source contact 208, a body region including regions 210 and 214, an interface 212 between regions 210 and 214, and a drain contact 216, according to one embodiment. The source metal region 202 may comprise aluminum or other conductive material. The gate insulation layer 204 may include a dielectric layer such as silicon dioxide or other dielectric material. The insulated gate 206 may comprise a polysilicon or aluminum gate surrounded by a dielectric layer such as silicon dioxide or other dielectric material. The source contact 208 may include a diffusion region, such as an N + diffusion region, in the body of the device cell 200. The source contact 208 may be diffused into another diffusion region, such as a P-base region for forming a channel region. Region 210 may include P-type pillars and region 214 may include an N-type epitaxial region. Regions 210 and 214 will support a reverse bias voltage when the device is in the off state. Interface 212 includes a PN junction between regions 210 and 214, wherein a depletion region is formed on each side of junction 212 when the device is in an off state. Drain contact 216 may be a metalized bottom surface of device cell 200, such as aluminum or an aluminum alloy. The body of device cell 200 may include an N-drift region in the epitaxial region above the N + substrate. Region 210 also includes a P + contact 203, which is shorted to source metal region 202 as shown.

Note that the single insulated gate 206 and the single source contact 208 are arranged asymmetrically with respect to the gate insulation layer 204. In various embodiments, both the insulated gate 206 and the source contact 208 have a length that is less than half the length of the gate insulation layer 204. In some embodiments, both the insulated gate 206 and the source contact 208 have a length that is less than one-third of the length of the gate insulation layer 204. In one embodiment, the length of the polysilicon gate 206 may be a minimum length, thereby reducing the corresponding gate-to-drain overlap.

Fig. 2B is a cross-sectional view of the SJ VDMOS device cell 200 of fig. 2A, including an indication of the vertical currents 222, 218 within the device in an active mode of operation. In order to simplify the drawing understanding, not all reference numerals are repeated in fig. 2B. The main path 222 of current through the body of the device cell 200 includes a single current branch 218, which current branch 218 flows up the main top surface of the device and through the channel to the single source contact 208. Thus, current branches 222 and 218 represent the same total vertical current through the device.

When comparing device cell 200 with exemplary device cell 100, gate charges Qg and Qgd are significantly reduced and a much lower FOM (rds (on) x Qg) is achieved. A somewhat higher specific rds (on) (rds (on) xaa is achieved, where AA is the active area of the SJ VDMOS device. Comparing the device cell 200 with the prior art device cell 100, only half of the source contact length is used (208vs 108A and 108B), and the single source contact 208 is positioned asymmetrically with respect to the gate insulation layer 204. Furthermore, the length of the polysilicon gate 206 is significantly reduced relative to the polysilicon gate 106. For example, in one embodiment, the length of the polysilicon gate 206 may be less than half the length of the polysilicon gate 106. In another embodiment, the length of the polysilicon gate 206 may be less than one third of the length of the polysilicon gate 106. In another embodiment, the length of the polysilicon gate may be a minimum gate length based on the critical dimensions of a given semiconductor manufacturing process.

Both device cells 100 and 200 have the same P + contacts 103 and 203, and as previously described, the P + contacts 103 and 203 are shorted to the corresponding source metal regions 102 and 202. Removing one of the N + contacts and/or reducing the gate polysilicon length in device cell 200 does not change the source-drain overlap, as demonstrated below with respect to the simulation and measurement results shown in fig. 6 and 7. With respect to device cell 200, the source metal coupled to P + contact 203 on the right hand side of the device cell is important even though no current flows in that side of the device cell. When the device is in a reverse biased state, the voltage on P + contact 203 sets the voltage above region 110 to ground. If P + contact 203 is removed on that side, region 110 will be electrically floating and the breakdown voltage of the device cell will be lower.

Fig. 3 is a plan view of the device cell 100 of fig. 1A and 1B. For simplicity, only the outline of a cell including two device cells 100A and 100B is shown, where the first device cell 100A includes a first source contact stripe 108A, a second source contact stripe 108B, and a polysilicon gate stripe 106 corresponding to the same features shown in fig. 1A and 1B. Gate strip 106 has a first edge 109A that overlaps an edge of source strip 108A and a second edge 109B that overlaps an edge of source strip 108B.

The second device cell 100B also includes a first source contact stripe 108A, a second source contact stripe 108B, and a polysilicon gate stripe 106 corresponding to the same features shown in fig. 1A and 1B. Gate strip 106 of second device 100B also has a first edge 109A that overlaps an edge of source strip 108A and a second edge 109B that overlaps an edge of source strip 108B.

The overall cell pitch ("2 × cell pitch") of both device cells 100A and 100B is shown in fig. 3, and also "L," which is the length of the polysilicon gate of each device, is shown.

Fig. 4 is a plan view of the device of fig. 2A and 2B. For simplicity, only the outline of a cell including two device cells 200A and 200B is shown, where the first device cell 200A includes a single source contact stripe 208 and polysilicon gate stripe 206 corresponding to the same features shown in fig. 2A and 2B. Gate strips 206 have first edges 209 that overlap the edges of source strips 208.

The second device cell 200B also includes a single source contact stripe 208 and polysilicon gate stripe 206 corresponding to the same features shown in fig. 2A and 2B. The gate strip 206 of the second device cell 200B also has a first edge 209 that overlaps an edge of the source strip 208.

The overall cell pitch ("2 × cell pitch") of both device cells 200A and 200B is shown in fig. 4, and also "L'", which is the length of the polysilicon gate of each device, is shown. The gate polysilicon length in the prior art device cells 100A and 100B of fig. 3 is "L", where the gate polysilicon length of the device 200 shown in fig. 4 is "L '", where L' < L in one embodiment.

Note that device cells 200A and 200B include "charge reduction strips" 210 that are not associated with steering the vertical current of the device, and do not include source strips 208 or gate strips 206. In prior art devices, the source implant is performed on both sides of the polysilicon gate, whereas in the devices shown in fig. 2A, 2B and 4, only a single-sided source implant is performed according to one embodiment. Thus, the "charge reduction strip" 210 does not include source injection.

Fig. 4 shows a cell layout of a power device, where each cell includes a source 208 and an overlapping polysilicon gate 206. The charge reduction strip 210 separates the cell 200A from the next cell 200B, which cell 200B also includes a source 208 and an overlapping polysilicon gate 206 in the same order.

Fig. 5 shows a cell layout of a power device, where each cell is horizontally "flipped" with respect to adjacent cells. For example, cell 300A includes a polysilicon gate 306 and a charge reduction strip 310, the polysilicon gate 306 including a portion 309 that overlaps a source 308. The cell 300A is separated from the next cell 300B, the cell 300B comprising a flipped configuration comprising in order: including source 308 and overlapping polysilicon gate 306 of overlapping portion 309, and charge reduction strip 310.

Fig. 5 is a plan view of the device of fig. 2A and 2B shown in an alternative layout configuration in accordance with another embodiment. For simplicity, only the outline of a cell including two devices 300A and 300B is shown, where the first device 300A includes a single source contact strip 308 and polysilicon gate strip 306 corresponding to the same features shown in fig. 2A and 2B. Gate strips 306 have first edges that overlap the edges of source strips 308. Cell 300A includes, in order from left to right, a gate strip 306 and a source strip 308.

The second cell 300B also includes a single source contact stripe 308 and polysilicon gate stripe 306 corresponding to the same features shown in fig. 2A and 2B. The gate strip 306 of the second cell 200B also has a first edge that overlaps the edge of the source strip 308. Cell 300B includes source stripe 308 and gate stripe 306 in order from left to right in an inverted configuration relative to cell 300A.

Although the layout configuration may use the cell layout shown in fig. 4 and 5 that repeats in the same orientation throughout the array of such cells, other layout configurations are possible. For example, other configurations in which multiple cells having the same source/gate or gate/source configuration (e.g., three, four, or more cells having the same source/gate configuration or gate/source configuration) are repeated in sequence may be used. Then, the macro cell (macrocell) may be specified to have three, four, or more repeating units. The macro-cells themselves may be repeated or flipped as desired in constructing the final power device. Those skilled in the art will recognize that many such repeating configurations may be implemented for power devices using the cells shown in fig. 4, 5, or a mixture of the cells shown in fig. 4 and 5. Flipping of the cells or macro-cells may be done in the X-axis and Y-axis directions as desired.

Fig. 6 is a table 600 of simulation results for multiple SJ VDMOS devices including different polysilicon gate lengths according to an embodiment. Four simulations were run for single source contact embodiments with gate lengths of 1 μm, 1.4 μm, 1.8 μm and 2.2 μm. For comparison, simulations were run for an exemplary device having two source contacts with a total gate length of 3.6 μm.

Table 600 is based on a 100V BV (breakdown voltage data sheet rating) SJ VDMOS platform that summarizes FOM related device parameters. When comparing single source stripe devices with the exemplary dual source stripe device, a significant reduction in Qg (gate charge Qg is defined as the charge from zero to the point where the drive voltage Vgs is equal to the actual gate voltage of the device), Ciss (the effective input capacitance seen by the gate drive circuit, Ciss Cgs + Cgd, where Cds is shorted), Crss (reverse transfer capacitance, Crss Cgd, also known as miller capacitance), and FOM (quality factor: rds (on) x Qg) is achieved. The output capacitance Coss is also shown in fig. 6. Since the P + source contact remains where the N + source contact has been removed, the simulated value of the output capacitance Coss is not significantly affected by the gate length and the number of source contacts. Furthermore, Coss does not change significantly since the drain side of the device cell is unchanged, so that the source-drain overlap is unchanged.

The FOM percentage reduction is a function of the polysilicon gate length. Within a particular range, the shorter the length, the greater the improvement in FOM. Although the polysilicon gate length may be reduced to a minimum amount, this may result in a corresponding increase in rds (on). Thus, in some embodiments, the polysilicon gate length is selected to be higher than the minimum length geometry provided by the fabrication technique used, since rds (on) will be at the boundary of the initial sharp increase, which will be more dominant than the Qg reduction, so the FOM will be the same or even higher, so it is not possible to push this parameter to the point where even small variations will severely affect the device parameters.

Fig. 7 is a table 700 of simulation and measurement results for a plurality of VDMOS devices including different polysilicon gate lengths according to an embodiment. Table 600 uses a 250V BV (breakdown voltage data table rating) platform that shows that significant reductions in Qg (gate charge), Qgs (gate source charge), Qgd (gate drain charge), and FOM (quality factor) are achieved when comparing single source stripe devices to conventional dual source stripe devices. The rds (on) resistance is also shown in fig. 7, which shows only a slight increase in resistance. Based on the measured data, a FOM improvement of 40% was achieved when the polysilicon length was reduced by half. Similar to the 100V case shown in fig. 5, further FOM reduction can be achieved by further shrinking the polysilicon gate length within the range allowed by the corresponding process window.

In contrast to the exemplary devices shown in fig. 1A, 1B, and 3, the device embodiments shown in fig. 2A, 2B, 4, and 5 include a source contact on only one side of the polysilicon gate. According to the embodiments shown and described herein, there is no source contact on the other side of the polysilicon gate. Thus, the polysilicon gate length is significantly reduced, which is used to ensure that the channel is on/off during switching, but with minimal overlap with the drain. (the N-epitaxial layer described above is also part of the drain side because it is connected to the drain by N-type doping.) thus, Qgd (gate drain charge) is greatly improved. Furthermore, there is no polysilicon over the P-body region on the side where the N + source contact is removed. Thus, when the device is turned on, there is no inversion layer. Therefore, the gate-source overlap is reduced to half, which results in half Qgs (gate-source charge). Thus, the total Qg (gate charge) is significantly reduced, and rds (on) is correspondingly increased by a small amount. Since rds (on) increases by a percentage less than the reduction in Qg, FOM is significantly reduced.

Thus, embodiments are described in which, with respect to the two source contact exemplary device, half of the source contact and corresponding channel are removed and the length of the polysilicon gate is reduced. Devices according to embodiments described herein exhibit improved performance and lower FOM due to the reduction in Qg more than the increase in rds (on).

The device embodiments may be used in small power converters because the circuits used in small power converters are typically single ended and subject to hard switching. Device embodiments may also be used in low voltage applications, particularly when switching losses are dominant (e.g., top-switches of buck converters). The device embodiments may also be used in combination with conventional devices, for example, for best efficiency, the device embodiments as top switches and the conventional devices as bottom switches (where conduction losses dominate).

When constructing a power device, multiple cells as shown in fig. 2A, 2B, 4 and 5 may be used in any orientation. However, multiple cells as shown in fig. 1A, 1B, and 3 may also be used in any orientation to create a power device having a mix of cell types. In one embodiment, this mix of cell types may be used to implement a particular FOM.

In one example, a device includes cells, where each cell includes: a body comprising a major top surface and a major bottom surface; a gate on a major surface of the device and having a first length; a gate insulating layer over the gate electrode and having a second length that is at least twice the first length; a source contact in the device body adjacent to the gate; a source metal layer over the gate insulating layer; and a drain on the major bottom surface of the cell. The device may comprise a plurality of substantially identical cells, wherein the gate is asymmetric with respect to the gate insulating layer, and wherein the source contact is asymmetric with respect to the gate insulating layer. The second length may be at least three times the first length. The gate may comprise a minimum length gate and may comprise a polysilicon gate.

In another example, a device includes cells, where each cell includes: a gate strip having a first edge and a second edge; a source strip extending along and overlapped by first edges of the gate strips; and a charge reduction strip extending along a second edge of the gate strip. The device may include a plurality of substantially identical cells, and further include forming a channel under the gate strips in the active mode of operation, and not forming a channel under the charge reduction strips in the active mode of operation. The length of the charge reduction strip is greater than the length of the source or gate strips. The gate strips may include minimum length gate strips and may include polysilicon gate strips.

One example method of fabricating a device cell includes: forming a gate having a first length on a main surface of the device unit; forming a gate insulating layer over the gate electrode, the gate insulating layer having a second length, the second length being at least twice the first length; forming a source contact adjacent to the gate; forming a source metal layer on the gate insulating layer; and forming a drain on the major bottom surface of the device cell. An example method includes forming a plurality of substantially identical cells, wherein a gate is formed asymmetrically with respect to a gate insulating layer, and wherein a source contact is formed asymmetrically with respect to the gate insulating layer. Forming the gate may include forming a minimum length gate and may include forming a polysilicon gate.

While the invention has been described with reference to illustrative embodiments, this description is not intended to be construed in a limiting sense. Various modifications and combinations of the illustrative embodiments, as well as other embodiments of the invention, will be apparent to persons skilled in the art upon reference to the description. It is therefore intended that the appended claims cover any such modifications or embodiments.