Semiconductor device

[0001] Field-effect controlled power switching devices such as a Metal Oxide Semiconductor Field Effect Transistor (MOSFET) or an Insulated Gate Bipolar Transistor (IGBT) have been used for various applications including but not limited to use as switches in power supplies and power converters. One example illustrating the use of MOSFETS in a dc to dc converter is given in FIG. 1.

[0002] The direction of current flow through the field-effect controlled devices operating as switches may be different in different operating cycles of power converters. In a “forward mode” of the field-effect controlled device, the pn-body diode at the body-drain junction of the field-effect controlled device is reversely biased and the resistance of the device can be controlled by the voltage applied to the gate electrode of the field-effect controlled device. In a “reversed mode” of the field-effect controlled device, the pn-body diode is forward biased. This results in a loss which is mainly determined by the product of current flow and voltage drop across the body diode. To minimize losses during reverse mode of the field-effect controlled device, i.e., maximize efficiency of the power supply or power converter, a shunting device, e.g., a diode, can be switched in parallel to the body diode of the field-effect-controlled switching device. Ideally, the shunting device should conduct no current when the body diode is reverse-biased and turn on at a lower voltage than the body diode when the body diode is forward-biased. To avoid unwanted inductivities and capacities associated with the required contacts and supply lines of additional devices, integrated power devices including e.g., a MOSFET and a diode have been proposed.

[0003] Commonly, mainly Schottky diodes have been used as integrated shunting devices. A Schottky diode is characterized by a low forward voltage drop of about 0.4 V at a given typical current, a low turn-on voltage of about 0.3 V, fast turn off, and nonconductance when the diode is reverse biased. For comparison, a silicon pn-diode has a forward voltage drop of about 0.9 V at given typical current and a turn-on voltage of about 0.6 V to 0.8 V. The losses during reverse biasing of a silicon MOSFET can, therefore, be reduced by connecting a Schottky-diode in parallel to the pn-body diode. However, to create a Schottky diode a metal-semiconductor barrier must be formed. In order to obtain proper electric characteristics for the Schottky diode, the metal used for the Schottky-contacts likely differs from the metal used for other structures such as Ohmic metal-semiconductor contacts. This can complicate the manufacture of the device. Further, the quality of a Schottky diode is usually affected by subsequent processes required for forming the MOSFET. In addition, Schottky diode rectifiers suffer from problems such as high leakage current and reverse power dissipation. Also, these problems usually increase with temperature and current thus causing reliability problems e.g., for power supply and power converter applications. Therefore, monolithically integrated power devices including Schottky barrier diodes can cause design problems.

[0004] For these and other reasons, there is a need for the present invention.

[0005] According to an embodiment, a semiconductor device is provided. The semiconductor device includes a source metallization, a first field-effect structure and a second field-effect structure. The first and second field-effect structures include a source region of a first conductivity type which is connected to the source metallization and a body region of a second conductivity type which is adjacent to the source region. The first field-effect structure further includes a first gate electrode and a first insulating region which is arranged at least between the first gate electrode and the body region. A first capacitance is formed between the first gate electrode and the body region. The second field-effect structure further includes a second gate electrode which is connected to the source metallization and a second insulating region which is arranged at least between the second gate electrode and the body region. A second capacitance is formed between the second gate electrode and the body region. The capacitance per unit area of the second capacitance is larger than the capacitance per unit area of the first capacitance.

[0006] According to another embodiment, a method for manufacturing a semiconductor device is provided. A semiconductor substrate of a first conductivity type is provided. At least a first trench and at least a second trench are formed in the semiconductor substrate. At least a lower portion of the walls of the first trench and a lower portion of the walls of the second trench are covered with a first oxide layer. A conductive region is formed at least in the lower portion of the first trench and at least in the lower portion of the second trench. A protecting region is formed on the second trench. A first insulating region is formed on the side walls in an upper portion of the first trench by a thermal oxidation process. During the thermal oxidation process the second trench is protected by the protecting region such that the semiconductor substrate forming the walls of the second trench is not oxidized during the thermal oxidation process. A second insulating region is formed on the side walls in an upper portion of the second trench. A first gate electrode and a second gate electrode are formed in the upper portion of the first and second trench, respectively. Source regions of the first conductivity type and a body region of a second conductivity type are formed such that the body region is adjacent to the source regions. A source metallization is formed, which is in contact to the source regions and the second gate electrode.

[0007] Further embodiments, modifications and improvements of the semiconductor device and the method will become more apparent from the following description and the appending claims.

[0031] In the following Detailed Description, reference is made to the accompanying drawings, which form a part hereof, and in which is showillustraten by way of illustration specific embodiments in which the invention may be practiced. In this regard, directional terminology, such as “top,” “bottom,” “front,” “back,” “leading,” “trailing,” etc., is used with reference to the orientation of the Figure(s) being described. Because components of embodiments of the present invention can be positioned in a number of different orientations, the directional terminology is used for purposes of illustration and is in no way limiting. It is to be understood that other embodiments may be utilized and structural or logical changes may be made without departing from the scope of the present invention. For example, features illustrated or described as part of one embodiment can be used on or in conjunction with other embodiments to yield yet a further embodiment. It is intended that the present invention includes such modifications and variations. The examples are described using specific language which should not be construed as limiting the scope of the appending claims. The drawings are not scaled and are for illustrative purposes only. For clarity, the same elements or manufacturing processes have been designated by the same references in the different drawings if not stated otherwise.

[0032] The terms “lateral” and “horizontal” as used in this specification intends to describe an orientation parallel to a first surface of a semiconductor substrate or body. This can be for instance the surface of a wafer or a die.

[0033] The term “vertical” as used in this specification intends to describe an orientation which is arranged perpendicular to the first surface of the semiconductor substrate or body.

[0034] In this specification, n-doped is referred to as first conductivity type while p-doped is referred to as second conductivity type. It goes without saying that the semiconductor devices can be formed with opposite doping relations so that the first conductivity type can be p-doped and the second conductivity type can be n-doped. Furthermore, some Figures illustrate relative doping concentrations by indicating “−” or “+” next to the doping type. For example, “n⁻” means a doping concentration which is less than the doping concentration of an “n”-doping region while an “n+”-doping region has a larger doping concentration than the “n”-doping region. Indicating the relative doping concentration does not, however, mean that doping regions of the same relative doping concentration have the same absolute doping concentration unless otherwise stated. For example, two different n+ regions can have different absolute doping concentrations. The same applies, for example, to an n+ and a p+ region.

[0035] Specific embodiments described in this specification pertain to, without being limited thereto, power semiconductor devices which are controlled by field-effect and particularly to unipolar devices such as MOSFETs, bipolar devices such as IGBTs and unipolar and bipolar devices having compensation structures such as Superjunction-MOSFETs.

[0036] The term “field-effect” as used in this specification intends to describe the electric field mediated formation of an “inversion channel” and/or control of conductivity and/or shape of the inversion channel in a semiconductor region of the second conductivity type. Typically, the semiconductor region of the second conductivity type is arranged between two semiconductor regions of the first conductivity type and a unipolar current path through a channel region between the two semiconductor regions of the first conductivity type is formed and/or controlled by the electric field. The conductivity type of the channel region is typically changed to the first conductivity type, i.e., inverted, for forming the unipolar current path between the two semiconductor regions of the first conductivity type.

[0037] In the context of the present specification, the semiconductor region of the second conductivity type in which an inversion channel can be formed and/or controlled by the field effect is also referred to as body region.

[0038] In the context of the present specification, the term “field-effect structure” intends to describe a structure which is formed in a semiconductor substrate or semiconductor device and has a gate electrode which is insulated at least from the body region by a dielectric region or dielectric layer. Examples of dielectric materials for forming a dielectric region or dielectric layer between the gate electrode and the body region include, without being limited thereto, silicon oxide (SiO₂), silicon nitride (Si₃N₄), silicon oxi-nitride (SiO_xN_y), zirconium oxide (ZrO₂), tantalum oxide (Ta₂O₅), titanium oxide (TiO₂) and hafnium oxide (HfO₂).

[0039] Above a threshold voltage V_th between the gate electrode and the body region, an inversion channel is formed and/or controlled due to the field-effect in a channel region of the body region adjoining the dielectric region or dielectric layer. The threshold voltage V_th typically refers to the minimum gate voltage necessary for the onset of a unipolar current flow between the two semiconductor regions of the first conductivity type, which form the source and the drain of a transistor.

[0040] In the context of the present specification, devices such as MOS-controlled diodes (MCDs), MOSFETs, IGBTs and devices having compensation structures such as Superjunction-MOSFETs as well as integrated devices with different field-effect structures are also referred to as field-effect structures.

[0041] In the context of the present specification, the term “MOS” (metal-oxide-semiconductor) should be understood as including the more general term “MIS” (metal-insulator-semiconductor). For example, the term MOSFET (metal-oxide-semiconductor field-effect transistor) should be understood to include FETs having a gate insulator that is not an oxide, i.e., the term MOSFET is used in the more general term meaning IGFET (insulated-gate field-effect transistor) and MISFET, respectively.

[0042] FIG. 2 illustrates an embodiment of a power semiconductor device 100 in a vertical cross-section. The semiconductor device 100 includes a semiconductor substrate 1 having a first surface 30 and a second surface 31 arranged opposite to the first surface 30. The semiconductor substrate 1 can be made of any semiconductor material suitable for manufacturing a semiconductor device. Examples of such materials include, without being limited thereto, elementary semiconductor materials such as silicon (Si), group IV compound semiconductor materials such as silicon carbide (SiC) or silicon germanium (SiGe), binary, ternary or quaternary III-V semiconductor materials such as gallium arsenide (GaAs), gallium phosphide (GaP), indium phosphide (InP), indium gallium phosphide (InGaPa) or indium gallium arsenide phosphide (InGaAsP), and binary or ternary II-VI semiconductor materials such as cadmium telluride (CdTe) and mercury cadmium telluride (HgCdTe) to name few. The above mentioned semiconductor materials are also referred to as homojunction semiconductor materials. When combining two different semiconductor materials a heterojunction semiconductor material is formed. Examples of heterojunction semiconductor materials include, without being limited thereto, silicon-silicon carbide (Si—SiC) and SiGe graded heterojunction semiconductor material. For power semiconductor applications currently mainly Si, SiC and Si—SiC materials are used.

[0043] The semiconductor substrate 1 can be a single bulk mono-crystalline material. It is also possible, that the semiconductor substrate 1 includes a bulk mono-crystalline material and at least one epitaxial layer formed thereon. Using epitaxial layers provides more freedom in tailoring the background doping of the material since the doping concentration can be adjusted during deposition of the epitaxial layer or layers.

[0044] Typically, the semiconductor substrate 1 is formed by providing a single bulk mono-crystalline body 1′ of a first conductivity type (n-doped) on which one or more single-crystalline layers 2 are deposited epitaxially. The epitaxial layer or layers 2 accommodates or accommodate an n-doped drift region 40, a p-doped body region or body regions 50 and an n-doped source region or source regions 80. During epitaxial deposition, the desired doping concentration of the drift region 40 can be adjusted by supplying an appropriate amount of dopant. Different thereto, the body region or regions 50 and the source region or regions 80 are typically formed in the epitaxially deposited drift region 40 by implantation. It would also be possible to form the body region 50 during epitaxial deposition by appropriately providing dopants of the second conductivity type (p-doped) in the desired concentration. The source region 80 can also be formed as a substantially continuous layer by implantation or during epitaxial deposition. If desired, the manufacturing can include separate epitaxial deposition processes with different dopants of varying concentration or with the same dopant but with varying concentration to form the respective functional regions. In some embodiments, the final doping concentration of the drift region 40 can vary to include doping profiles having at least one minimum or at least one maximum or having an increasing or decreasing doping concentration from a drain region 41 to the body region 50.

[0045] In other embodiments, a substrate wafer or die having the desired background doping concentration of the drift region 40 is provided. The substrate wafer is suitably thinned and body region 50 and source region 80 are formed by implantation at the first surface 30. If desired, the substrate wafer can be further thinned at the second surface 31 and the drain region 41 is formed by implantation at the second surface 31. It would also be possible to thin the substrate wafer after implanting source and body regions 80, 50 only. By using this approach, an expensive epitaxial deposition can be avoided.

[0046] The semiconductor substrate 1 of FIG. 2 includes a common drift region 40 and a spaced apart source region 80 both of the n-conductivity type. Typically, the source region 80 is in electrical contact with a common source metallization 60, and the doping concentration of the source region 80 is higher than the doping concentration of the drift region 40 as indicated by the symbols “n+” and “n”. Between the drift region 40 and the source region 80 a p-doped body region 50 is arranged and respective pn-junctions between the source region 80 and the body region 50 and between the body region 50 and the drift region 40 are formed. At least two first trenches 10 and at least a second trench 20 laterally arranged between the two first trenches 10 extend from the source region 80 through the body region 50 partially into the drift region 40. Typically, the first and second trenches 10 and 20 extend in a direction perpendicular to the illustrated cross-section. The trenches may, however, have any shape and can e.g., be formed as stripes. Typically, the trenches have, in a vertical cross-section, a width of about 0.5 μm to about 2 μm and a lateral distance of about 0.5 μm to about 2 μm.

[0047] The sidewalls and the bottom walls of the first trenches 10 and the second trench 20 illustrated in FIG. 2 are covered with a first insulating region 12 and a second insulating region 22, respectively. The insulated first and second trenches 10 and 20 are filled with a first conductive region 11 forming a first gate electrode 11 and second conductive region 21 forming a second gate electrode 21, respectively. The material of the first and second gate electrode 11 and 21 may be a metal such as Ti, W and Co or a material with metallic or near metallic properties with respect to electric conductivity such as highly doped n-type or p-type poly-Si, TiN or an electrically conductive silicide such as WSi₂. Due to their metallic properties each of the first and second gate electrodes 11 and 21 forms with the respective first and second insulating region 12 and 22 and the adjoining body region 50 a metal-insulator-semiconductor (MIS) structure.

[0048] In the context of the present specification, the term “gate electrode” intends to describe an electrode which is situated next to, and insulated from a body region 50, i.e., “gate electrodes” may also be those electrodes which are not at gate potential. The gate electrodes may be formed on top of the semiconductor substrate 1 or between mesa regions. In the context of the present specification, the term “mesa” or “mesa region” intends to describe the semiconductor region between two adjacent trenches extending into the semiconductor substrate in a vertical cross-section.

[0049] The second gate electrode 21 is in contact with a source metallization 60 which is also in contact with the source region 80 and the body region 50.

[0050] In the context of the present specification, the terms “in Ohmic contact”, “in electric contact”, “in contact” and “electrically connected” intend to describe that there is an Ohmic electric connection or Ohmic current path between two regions, portion or parts of a semiconductor devices, in particular a connection of low Ohmic resistance, even if no voltages are applied to the semiconductor device. An Ohmic electric connection is characterized by a linear and symmetric current-voltage (I-V) curve.

[0051] Due to the body diodes 15 formed by the pn-junctions between the body region 50 and the common drift region 40, the source metallization 60 and the drift region 40 are e.g., not in contact.

[0052] The first gate electrodes 11 are in contact with a gate metallization (not illustrated in FIG. 2). Further, the drift region 40 is in Ohmic contact with a common drain metallization 42 on the second surface 31 of the semiconductor device 100, wherein for better contact, a highly n-doped common drain region 41 can be arranged between the common drift region 40 and the common drain metallization 42.

[0053] In the cross-sectional view, the device 100 has separated body regions 50 and separated source regions 80. The source regions 80 adjoining a first trench 10 and a second trench 20 may also be referred to as first source regions and second source regions, respectively. However, the source regions 80 and/or the body regions 50 may also be simply connected at least in respective pairs. The electric contact between the source metallization 60 and the body region 50 may e.g., only be realized in certain portions of the semiconductor device 100. In this case the illustrated source regions 80 between two adjacent trenches are simply connected. Typically, even actually separated body regions 50 are in electric contact with each other. Further, even actually separated source regions 80 are typically in electric contact with each other too. For clarity reasons, apparently and actually separated body and source regions are labeled with the same respective reference sign.

[0054] According to a first embodiment, the capacitance per unit area C2 between the second gate electrode 21 and the body region 50, in the following also referred to as second capacitance per unit area, is larger than the capacitance per unit area C1, in the following also referred to as first capacitance per unit area, between the first gate electrode 11 and the body region 50. Typically, inversion channels can be formed along the first and second insulating region 12 and 22 in the body region 50. Due to the different capacitances per unit area between the body region 50 and the respective gate electrodes, the voltage differences between the body region 50 and the respective electrode, which is required to form the inversion channel, is typically lower for the second field-effect structure.

[0055] According to another embodiment, the permittivity of the second insulating region 22 is higher than the permittivity of the first insulating region 12 at least between the body region 50 and the respective gate electrodes 11 and 12. Thereby, the second capacitance per unit area C2 can be larger than the first capacitance per unit area C1 even at same geometry of the first and second trench 10 and 11. For example, the first insulating region 12 is made of SiO₂, Si₃N₄ or SiO_xN_y, whereas the second insulating region 22 is made of HfO₂. In another example, the first and second insulating region 12 and 22 are made of SiO₂ and Si₃N₄, respectively. The first and second insulating regions 12 and 22 may also include several layers of different materials. These layers should be chosen such that the second capacitance per unit area C2 is larger than the first capacitance per unit area C1.

[0056] According to yet another embodiment, the first gate electrode 11, the first insulating region 12, the source region 80 in contact with the source metallization 60, the body region 50 and the drift region 40 in contact with the drain metallization 42 forms a first field-effect structure, namely a MOSFET.

[0057] If the voltage V_GS between the gate metallization and source metallization 60 exceeds a threshold value, an n-type inversion channel 51 is formed along the first insulating region 21 in the body region 50 as indicated in FIG. 3 illustrating a similar semiconductor device 100 as FIG. 2 but in a different cross-section, in which the second gate electrode 21 is also spaced apart from the source metallization 60 by a dielectric portion 70. However, the second gate electrode 21 is in contact with the source metallization 60 in other portions of the device 100. In other words, there is at least a second cross-section through the semiconductor device 100 of FIG. 3 illustrating that the source metallization 60 is connected to the second gate electrode 21. This applies to all Figures of the present specification in which the contact between the second gate electrode 21 and the source metallization 60 is not illustrated.

[0058] According to yet another embodiment, the second gate electrode 21, which is in contact with the source region 80 and the source metallization 60, the second insulating region 22, the body region 50 and the drift region 40 in contact with the drain metallization 42 form a second field-effect structure, which is in the following referred to as MOS-gated diode (MGD). The term “MOS-gated diode” or “MGD” as used in this specification intends to describe a MOSFET structure with shorted gate electrode and source electrode, i.e., a MGD is a two terminal field-effect structure. Further, the body region 50 of the MGD is typically in contact with the source electrode 60. Typically, the MGD is connected in parallel to the body diodes 15 formed between the body region 50 and the drain region 40.

[0059] In other words, embodiments as described herein include an integrated semiconductor device which has a body diode 15 formed between a body region 50 and a common drift region 40, a first field-effect structure and a second field effect structure which is typically a MGD. The first field-effect structure and the second field effect structure are typically connected to a first common metallization and a second common metallization. Typically, the first common metallization is electrically connected to the source regions 80 of the first and second field effect structures. This metallization is therefore typically referred to as source metallization 60. The body region 50 is typically also connected to the first common metallization. The second common metallization is typically in electrical contact with the common drift region 40. The total current between the two common metallization may flow in either direction through the integrated semiconductor device.

[0060] In a “forward mode” of the semiconductor device, in which the body diode 15 is reversely biased, the first field-effect structure can control the resistance of the semiconductor device by the field-effect. Therefore, the first field-effect structure is also referred to as controllable field-effect structure. To control the resistance, an appropriate voltage difference between the first common metallization and an insulated gate electrode 11 of the first field-effect structure is applied or changed as known to those skilled in the art. Thereby, an inversion channel 51 within the body region 50 can be formed and/or modified and the current blocking body diode 15 can be bypassed. At given voltage difference between the first and second common metallization the total current flowing through the semiconductor device can by controlled in this way.

[0061] In a “reverse mode” or “backward mode” of the semiconductor device, the body diode 15 is forwardly biased. Furthermore, since the body region 50 and the source region 80 are shorted in many embodiments, a current can flow through the device in backward mode. Further, the insulated gate electrode 21 of the second field-effect structure is shorted with the source metallization 60. Thus the current cannot be controlled by applying a control voltage to the second field-effect structure. However, an inversion channel can also be formed in the reversed mode under specific conditions. Generally, the forming of an inversion channel in the channel region of a p-type body region next to an insulated gate electrode requires a positive voltage difference between the insulated gate electrode and the body region V_GB>0. Even if the body contact and the insulated gate electrode are electrically connected, a positive voltage difference can occur depending on the built-in potential between the source region 80 and the body region 50, the voltage drop due to the current flow from the source region 80 to the drain region 41, and on the work function differences between the gate material and the material of the body region 50.

[0062] Due to the resistivity of the body region 50, any current flow during reverse mode reduces the voltage along the current path in the body region 50 to values which are lower than the voltage V_S applied to the source metallization 60. This typically results in a lower potential of the body region next to the insulated gate electrode. Therefore, the voltage difference V_GB increases typically with the current and current density, respectively.

[0063] In certain embodiments the second field-effect structure (MGD) is designed such that the total current through the integrated semiconductor device in reverse mode is, above an average current flow density threshold, typically dominated by a unipolar current flowing via an inversion channel 52 along the insulating gate electrode 21. Typically, this reduces the electric losses of the integrated semiconductor device compared to the case of a total current flow across the pn-junctions of the body diode 15 during reverse mode.

[0064] Further, not the electric potential but the quasi-Fermi level of the electrons (and that of the holes) is typically equalized between the metallic gate electrode, the metallic source electrode and the metallic body contact, when the contacts are short-circuited. Therefore, a positive potential difference V_GB between a gate electrode, in particular the second gate electrode 21, and the body region 50 can be formed even without applying an external voltage or current to the semiconductor device 100. The gate potential V_G reads:
V_G=E_g(material_body)/2+χ(material_body)−WF(material_{gate electrode})
with work function WF, electron affinity χ and band gap E_G.
For a monocristalline silicon body and highly phosphorous doped polycrystalline silicon (poly-Si) electrodes, the gate potential V_G typically amounts to about V_G=0.56 V+4.17 V−4.35 V=0.37 V.

[0065] In the context of the present specification, the term “work function” intends to describe the minimum energy (usually measured in electron volts) needed to remove an electron from a solid to a point outside the solid surface. This corresponds for metals to the energy needed to move an electron from the Fermi energy level, which lies within the conduction band, into vacuum. For a semiconductor material or an insulator the work function can be defined as the sum of the electron affinity χ and half of the band-gap, i.e., the minimum energy needed to move an electron from the intrinsic Fermi level into vacuum.

[0066] Gate electrode materials having a work function which is lower than the above given value of 4.35 V for highly phosphorous doped polysilicon will produce even higher positive V_GB than 0.37 V. In some embodiments, the work functions of the first and second gate electrode 11 and 21 are different. Typically, the work function of the second gate electrode 21 is smaller than the work function of the first gate electrode 11. For example, the first gate electrode 11 is made of highly doped poly-Si, and the second gate electrode is made of TiN, TaN or Co. Typically, the electron affinity of the body region 50 is also smaller than the work function of the first gate electrode 11. For example, the first gate electrode 11 and the body region 50 are made of highly doped poly-Si and of Si, respectively.

[0067] If the voltage difference V_GB between the insulated gate electrode and the body region is larger than a threshold voltage V_th, an inversion channel is formed along the insulated gate electrode in the body region 50.

[0068] Generally, the threshold voltage V_th of a field-effect structure reduces with increasing gate capacitance per unit area and decreasing doping concentration of the body region 50. This applies both for a MOSFET structure during “threshold connection” in forward mode (V_GS=V_DS>0) and a MOSFET structure in reverse mode (or “reverse threshold connection”, V_DG=V_DS<0) with the voltage differences V_GS, V_DG and V_DS between gate and source, drain and gate, and drain and source, respectively. During “reverse threshold connection” of the MOSFET, the drain is used as electron source and the source is used as electron drain. In addition to the electron transport through the inversion channel of the MOSFET, the current of the reverse bipolar transistor in the mesa and the hole current across the pn-body diode typically contribute to the total current in reverse mode. Therefore, the threshold voltage V_th of the MGD is typically lower than the threshold voltage V_th of the MOSFET even at same capacitance per unity area between the body region 50 and the respective gate electrode.

[0069] Further, only a weak inversion channel or weak inversion layer 52, which has a charge carrier concentration of about 10¹⁷ cm⁻³ to about 10¹⁸ cm⁻³, is typically formed along the second insulating region 22 in the body region 50 of the MGD.

[0070] Since the second gate electrode 21 is connected to the source metallization 60 in the FIGS. 2 and 3, the lower threshold voltage V_th of the MGD is typically not reflected in the gate characteristic of the MOSFET with integrated MGDs. Further, the maximum rated gate voltage of the MOSFET does not result in a lower limit for the gate thickness of the MGD.

[0071] The inversion channel is typically only formed along the second insulating region 22 during reverse mode. This is because the second field-effect structure (MGD) has a higher capacitance per unit area between its gate electrode and the body region than the first field-effect structure (MOSFET).

[0072] The voltage drop across the semiconductor device 100 can, depending on the current density and the properties of the MGDs, typically be reduced from about 0.9 V of the body diode 15 to values below 0.5 V during reverse mode of the integrated MOSFET 100. Thereby, the losses are reduced in this mode. The use of MOSFETs with integrated MGD 100 in a typical converter can therefore increase the converter efficiency. This is explained in more detail with reference to FIG. 1.

[0073] FIG. 1 illustrates a circuit diagram of a typical step-down dc to dc converter, i.e., a buck converter, using MOSFETS. An input voltage U_in is stepped down to a lower output voltage U_out. The topology of the illustrated circuitry is widely used, e.g., on computer mainbords, to convert a typical input voltage U_in of 12 V provided by the mains adapter to the required voltages of e.g., about 1.2 V to about 3.3 V of the consumers of the mainboard such as a CPU, a GPU, a DSP, a DRAM and driver chips. The buck converter has four operating phases which are controlled by a driver IC 95. In a first phase the highside MOSFET switch 96 is switched on and the two lowside MOSFET switches 97 are switched off. This causes a linear current increase through the inductor 98 charging the capacitance 99. If the output voltage U_out exceeds a certain threshold, the driver IC 95 switches the MOSFET 96 off which initiates the second phase. Now the load current flows in the freewheel circuit formed by the inductor 96, the capacitance 97 and the body diodes of the two MOSFETs 97. In this phase the MOSFETs 97 are in reverse mode, and the losses are mainly caused by the body diodes which are now biased in forward direction. Typically, the forward voltage drop of the body diode of a silicon MOSFET is for typical currents about 0.9 V or even larger. After a dead time the MOSFETs 97 are switched on by the driver IC 95 to reduce losses (third phase). If the output voltage falls below a limit the MOSFETs 97 are switched off again (fourth phase) prior to returning to the first phase. To minimize the losses of the buck converter, MOSFETs 100 with integrated second field effect devices having a low voltage drop if the body diode is switched in forward direction can be used. This applies also to other type of converters such as step-up converters and single ended primary inductance converters.

[0074] Except for the capacitances per unit area, the technical features can be optimized independently for the first and second field-effect structures. Examples of such features include but are not limited to the leakage current, the blocking ability, the quality of Ohmic contacts and related temperature dependencies.

[0075] Further, different threshold voltages for the MOSFETs which are higher than the threshold voltage of the second field effect structure (MGD) on a single integrated circuit may be required. This can e.g., be obtained by selectively providing channel implants for the first field effect structures forming the respective transistors. Additional channel implants, i.e., the doping of the channel region 51 to adjust the threshold voltage of the first field effect structures, may be used for those MOSFETS which have different threshold voltage requirements V_th.

[0076] Further, the concept of integrating a first field-effect structure having a first capacitance per unit area between its gate electrode and a body region and a second field-effect structure, which includes a shorted gate electrode and source electrode and has a capacitance per unit area between its gate electrode and the body region which is higher than the first capacitance per unit area, is not limited to the illustrated vertical field-effect structures with gate electrodes arranged in trenches as illustrated in FIGS. 2 and 3 (VMOSFET, UMOSFET). In further embodiments the principles disclosed herein are also used in lateral devices such as a lateral MOSFET and in planar vertical devices, i.e., devices with non-buried gate electrode, such as a DMOSFET.

[0077] In other words, the semiconductor device 100 includes a source metallization 60 in contact with a source region 80 of a first conductivity type, a drain region 41 of the first conductivity type and a body region 50 of a second conductivity type. The body region 50 respectively adjoins the source region 80 and the drift region 40. The semiconductor device 100 further includes a first field-effect structure with a first gate electrode 11 and a first capacitance per unit area C1 between the first gate electrode 11 and the body region 50, and a second field-effect structure with a second gate electrode 21 and a second capacitance per unit area C2 between the second gate electrode 21 and the body region 50 which is larger than the first capacitance per unit area C1.

[0078] According to certain embodiments, the semiconductor device is a power-semiconductor device which includes a plurality of monolithically integrated first and second field-effect structures. In other embodiments, the semiconductor device 100 includes only one first and/or one second field-effect structure.

[0079] Referring again to FIG. 3 still a further embodiment will be explained. Accordingly, the thickness d₂ of the second insulating region 22 between the second gate electrode 21 and the body region 50, in the following also referred to as second thickness, is smaller than the thickness d₁ of the first insulating region 12 between the first gate electrode 11 and the body region 50. In the following the thickness d₁ is also referred to as first thickness. Thereby, the second capacitance per unit area C2 can be larger than the first capacitance per unit area C1 even if the same electrically insulating material is used for forming the first and second insulating region 12 and 22.

[0080] For example, for a silicon oxide as gate insulating material, the first thickness d1 is typically in a range between about 10 nm and about 100 nm.

[0081] The second thickness d₂ can be significantly smaller, e.g., by a factor of two or more than a typical thickness of a silicon oxide layer as gate insulator of about 40 nm to 60 nm in standard power MOSFETs. In certain embodiments, the second thickness d₂ is smaller than about 8 nm. The second thickness d₂ may be smaller than 6 nm or 4 nm and may even be smaller than 1 nm.

[0082] Typically, the second thickness d₂ is smaller than the maximum thickness of the second insulating region 22 between the second gate electrode 21 and the common drift region 40. Further, the first thickness d1 is typically smaller than the maximum thickness of the first insulating region 22 between the first gate electrode 11 and the common drift region 40.

[0083] FIG. 4 illustrates a vertical cross-section of a semiconductor device 100 according to further embodiments. The illustrated semiconductor device 100 differs from the one illustrated in FIG. 3 in that it includes two second trenches 20 next to each other. Further, in each of the first and second trenches 10 and 20 field plates 16 and 26 were formed below the respective gate electrodes 11 and 21. The two second trenches 20 are spaced to respective neighboring first trenches 10 by a mesa region of a first lateral distance p₁. In addition, the two second trenches 20 are spaced apart by a mesa region of a second lateral distance p₂. In certain embodiments, the first lateral distance p₁ is larger than the second lateral distance p₂ and/or the second field plates 26 extends vertically deeper into the common drift region 40 than the first field plates 16. Since the second gate electrodes 21 and the first and second field plates 16 and 26 are on source potential, the drift region 40 in the mesa between the two second trenches 20 is screened against high electric field strength in forward mode. Consequently, the second field effect structure, i.e., the integrated MGD, is typically better protected against Avalanche breakdown than the first field effect structure.

[0084] Due to the arrangement of the first and second trenches 10 and 20, there are four first body sub-regions 50a which adjoin the first insulating region 12 and one second body sub-region 50b which does not adjoin the first insulating region 12 but adjoins the second insulating regions 22 of the neighboring second trenches 20. In some embodiments the second body sub-region 50b has lower doping concentration than the first body sub-regions 50a. This will typically further reduce the threshold voltage V_th for forming the inversion channel of the second field effect structure and hence the voltage drop during reverse mode.

[0085] The semiconductor device 100 illustrated in FIG. 5 differs from the one illustrated in FIG. 3 in the geometry of the insulating regions 12 and 22 in a lower portion of the first and second trenches 10 and 20, respectively. Typically, both insulating regions include two respective insulating portions, a first and a second insulating portion 12a and 22a between the body region 50a or 50b and the respective gate electrode 11 and 22 and first and second insulating bottom portion 12c and 22c filling at least the space between the bottom of the trenches 10 and 20 and the respective gate electrodes 11 and 21. In some embodiments, the lateral and/or vertical thickness of the insulating bottom portions 12c and 22c below the respective gate electrodes exceeds the respective thickness of the insulating portions 12a and 22a in a vertical cross-section. Thereby, the field strength in the bottom portions 12c and 22c can be reduced. Typically, the lateral and/or vertical thickness of the first and second insulating bottom portions 12c and 22c is in a range of about 50 nm to about 300 nm.

[0086] With respect to FIG. 6 further embodiments will be explained. The semiconductor device 100 illustrated in a vertical cross-section includes an n-type source region 80 in contact with a common source metallization 60. The source region 80 adjoins a p-type body region 50 which adjoins a common n-type drift region 40. Between the body region 50 and the drift region 40 a body diode (not illustrated) is formed. Within the drift region 40 a third conductive region 25 of the p-type is buried. Typically the doping concentration of the third conductive region 25 is higher than the doping concentration of the body region 50. Further, the third conductive region 25 and the body region 50 are spaced apart from each other. Due to the formed pn-junction between the third conductive region 25 and the drift region 40, a space charge region or layer is typically formed next to the pn-junction. A second trench 20 extends from the source region 80 through the body region 50 and at least partially into the drift region 40. The second trench 20 adjoins the third conductive region 25 and includes an insulating layer 22 and a conductive plug 21, which forms an Ohmic connection between the source metallization 60 and the third conductive region 25. The insulating layer 22 is only arranged on the side walls of the second trench 20 and insulates the conductive plug 21 from the body region 50 and the source region 50. The body region 50 may be connected to the source metallization 60.

[0087] In some embodiments as described herein, the conductive plug 21, the insulating layer 22 and the body region 50 form a second field effect structure which is typically a MGD having a second capacitance per unit area C2 between the conductive plug 21 forming a second gate electrode 21 and the body region 50.

[0088] In certain embodiments, the semiconductor device 100 further includes at least one first trench 10 which extends from the source region 80 through the body region 50 partially into the drift region 40. In FIG. 6 two first trenches 10 are exemplarily illustrated. The side walls and the bottom walls of the first trenches 10 are covered with a first insulating layer 12. The insulated first trenches 10 are filled with first conductive regions forming first gate electrodes 11.

[0089] In some embodiments as described herein, the second and first trenches can also be described as a trench and a further trench, respectively. In this case, the second field-effect structure and the first field-effect structure form a field-effect structure and a further field-effect structure, respectively.

[0090] Typically the capacitance C1 per unit area between the first gate electrode 11 and the body region 50 is lower than the second capacitance per unit area C2. This can again be achieved by choosing an appropriate effective thickness and/or permittivity of the first insulating region 12 and the insulating layer 22. In addition to the common source metallization 60, the semiconductor device 100 typically includes a common drain metallization 42 and a common gate metallization (not illustrated) in electrical contact with the first gate electrodes 11 so that the device 100 can be operated as a three-terminal MOSFET. As the MOSFET 100 includes a MGD which is connected in parallel to the body diode, the integrated MOSFET 100 has typically a lower voltage drop during reverse mode compared to standard MOSFETs. This favors the use of the integrated MOSFET 100 as a low-side MOSFET 97 in a converter as illustrated in FIG. 1.

[0091] In some embodiments, the first trench 10 further includes in the lower portion a conductive field plate 16 in contact with the source metallization 60 to allow a higher doping concentration and/or thinner drift region 40 while keeping the breakdown voltage substantially constant. The field plates 16 and the third conductive region 25 screen the body region 50 during forward mode. Further, the third conductive region 25 can carry an Avalanche current. Therefore, the body region 50 may also be floating.

[0092] FIG. 7 illustrates, in a vertical cross-section, a similar MOSFET with integrated MGDs 100 as illustrated in FIG. 6. In addition, the first and second inversion channels 51 and 52 are illustrated, which can be formed in the body region 50 by the field-effect such that they extend from the source region 80 to the drift region 40. For clarity reasons not all inversion channels of the semiconductor device 100 are labeled with the respective reference signs. In certain embodiments, the doping concentration of the channel regions 52 is lower than the doping concentration of the remaining part of the body region 50 to further reduce the threshold voltage of the second inversion channels 52. Due to the formed inversion channels 52 of the MGD during reverse mode, the amount of stored minority carriers (reverse recovery charge) is typically also reduced compared to standard MOSFETs. A reduction of the stored charge generally results in a reduction of the current peak during commutation. Thus, the switching behaviour of the MOSFET with integrated MGDs 100 can be improved compared to standard MOSFETs. Accordingly, the MOSFET with integrated MGDs 100 can also be used as low-side switch with improved switching behaviour in a converter circuit arrangement.

[0093] Further, first doped regions 27, second doped regions 28 and third doped regions 29 are illustrated in the cross-section of FIG. 7. The first doped regions 27 adjoin the source metallization 60, the source region 80, the body region 50 and the insulating layer or second insulating region 12. The second doped regions 28 adjoin the third conductive region 62, the body region 50, the common drift region 40, the insulating layer 12 and a respective third doped region 29. The third doped regions 29 are arranged between the body region 50 and the drift region 40 on both sides of each conductive plug 21. In certain embodiments, each of the first, second and third doped regions 27, 28 and 29 are regions of the first conductivity type, i.e., n-doped regions, having a doping concentration which is typically higher than the doping concentration of the drift region 40. Thereby, the length L₂ of the second inversion channel 52, which can be formed in the body region 50 along the second insulating region 22, can be tailored independently from the length L₁ of the first inversion channel 51, which can be formed in the body region 50 along the insulating layer 12. According to another embodiment, the length L₂ is smaller than the length L₁. Thereby, the electric resistance of the second inversion channel 52 can be further reduced. This results in even lower losses of the MOSFET with integrated MGDs 100 during reverse mode as desirable for many applications e.g., as lowside MOSFET 97 in the converter of FIG. 1.

[0094] In certain embodiments, a conductive contact layer 62 is arranged between the conductive plug 21 and the third conductive region 25 to improve the electric contact and to reduce the resistance between the source metallization 60 and the third conductive region 25. Typically, the contact layer 62 has a metallic or near metallic conductivity. For example, the contact layer 62 can be made of a metal, a silicide or Ti/TiN for improving the contact between a poly-Si plug 21 and a p-type third conductive region 25 made of silicon.

[0095] In another embodiment, some of the first doped regions 27a are of the p-type as illustrated in FIG. 8, which also illustrates a cross-section through a MOSFET with integrated MGDs 100. If the device 100 is not designed to have floating body regions 50, the p-type first doped regions 27a can be used for electrically connecting the body region 50 and the source metallization 60. The cross-section of FIG. 8 may also correspond to a further cross-section of the MOSFET with integrated MGDs 100 of FIG. 7. In other words, the shortening of the vertical extension of the second inversion channel and the electrical connecting of the body region 50 may be done in different portions of the semiconductor device 100.

[0096] With respect to the FIGS. 9-13 an embodiment of a method for manufacturing a MOSFET with integrated MGDs 100 is explained. FIG. 9 illustrates a vertical cross-section of the semiconductor device 100 after providing a semiconductor substrate which includes an n-type common drain region 41 and an n-type common drift region 40 and after further processes including forming first and second trenches 10 and 20, forming p-type body regions 50 and n-type source regions 80 and forming dielectric portions 70. In each of the first trenches 10 a field plate 16, a gate electrode 11 and an insulating region 12 were formed. Further, the second trenches 20 were etched through the source region 80 and the body region 50 partially into the common drift region 40. All these processes were performed using standard processes for forming vertical trench MOSFETS known to those skilled in the art.

[0097] Subsequently, an insulating layer 22 is arranged on the side walls and the bottom walls of the second trenches 20. This can be done by a thermal oxidation of the semiconductor substrate and/or by deposition of an insulating material. In some embodiments the thickness of the insulating layer 22 between the mesa and the recess of the second trenches 20 is smaller than the thickness of the first insulating region 12 between the body regions 50 and the first gate electrode 11. In certain embodiments the permittivity of the insulating layer 22 is higher than the permittivity of the first insulating region 12. FIG. 10 illustrates the semiconductor device 100 after a subsequent ion implantation process for forming p-type third conductive regions 25 in the drift region 40. The third conductive regions 25 adjoin the insulating layer 22 on the bottom of the second trench 20.

[0098] Thereafter, an anisotropic etching process is carried out to remove the insulating layer 22 on the bottom of the second trenches 20 as illustrated in FIG. 11.

[0099] Subsequently, a conductive material such as highly doped poly-Si is deposited in the second trench 20 for forming a conductive plug 21. The dielectric layer 22 and the conductive plug are etched back in an upper portion of the second trenches 20 to expose the source regions 80. This results in a structure as illustrated in FIG. 12.

[0100] Alternatively, the third conductive regions 25 can be formed after etching the insulating layer 22 on the bottom of the second trench 20 and filling the second trench 20 with poly-Si, e.g., by diffusion of boron out of the deposited poly-Si.

[0101] Finally, a common source metallization 60 and a common gate metallization (not illustrated) are formed on the top side and a common drain metallization 42 is formed on the bottom side of the semiconductor device 100 as illustrated in FIG. 13.

[0102] Since standard processes are used prior to etching the second trenches 20 the pitch and/or the lateral distance between two first trenches 10 next to each other has typically not to be increased compared to standard MOSFETS without integrated MGDs. Still the voltage drop during reverse mode (reversed current flow) can be significantly reduced as will be explained with respect to FIG. 14.

[0103] FIG. 14A illustrates, within the rectangular section 5, the current lines 19 of an integrated MOSFET as illustrated in FIG. 13 with integrated MGDs according to a numerical simulation. The insulating layer 22 is too thin (5 nm) to be clearly visible. For comparison the current lines 19 during reversed mode of the standard MOSFET with the same pitch is given in FIG. 14B. As can be seen the current in FIG. 14A is dominated by an electron current flowing from the source region 80 through the inversion channel 52 (not labelled) in the body region 50 to and through the drift region 40. In contrast thereto, the current in the standard MOSFET is bipolar under the same condition due the current flow across the body diode. As a result of the additionally formed inversion channel 52, the voltage drop across the MOSFET with integrated MGDs is only half as large as for the standard MOSFET in a wide current range in reverse mode. The corresponding current density-voltage-characteristics of FIGS. 14A and 14B are plotted in FIG. 14C as curves A and B, respectively.

[0104] Field plates may also additionally be incorporated in semiconductor devices as illustrated in FIG. 4. This is further illustrated in FIG. 15 illustrating in a vertical cross-section a section of a power-MOSFET 100 with a plurality of integrated MOSFETS and MGDs. Each of the illustrated first and second trenches 10 and 20 includes in its lower portion a field plate 16 and 26, respectively. The field plates 16 and 26 are connected to the source metallization as indicated by the reference sign “S”. The first gate electrodes 11 are connected to a not illustrated gate metallization as indicated by the reference sign “G”. For sake of clarity, only the first two trenches from the left of FIG. 15 are fully designated with reference signs. A more detailed section of the structure is given below in FIG. 17.

[0105] According to a further embodiment, the plurality of first field-effect structures (MOSFETS) and second field-effect structures (MGDs) are arranged in a regular pattern. Typically, this regular pattern at least extends over the major portion of the semiconductor device 100. The border area of the device may, however, deviate from the pattern e.g., to compensate boundary effects. In FIG. 15 every fourth field-effect structure is a MGD. As can be seen from the additionally plotted electron current lines 19 during normal MOSFET operation of the MOSFET 100, i.e., during forward mode in which the electrons flow from source metallization 60 through the source region 80, the inversion channels 51 in the body region 50 and the drift region 40 to the drain metallization 42, every mesa contributes to the total current flow. A closer inspection of the current lines 19 reveals that the integration of MGDs increases the resistance in forward mode R_on only by 22% which is lower than the expected increase of 33%.

[0106] On the other hand, during reversed current (reverse mode) the electron current flows from the drain metallization 42 through the drift region 40, the inversion channel 52 in the body region 50 next to the only 5 nm thick gate insulation 22 and the source region 80 to the source metallization 60. This is illustrated in FIG. 16 illustrating the same MOSFET as in FIG. 15 but during reverse mode. Due to the lower voltage drop across the inversion channel 52 compared to the body diode of the MOSFET, the losses during reverse mode can be reduced significantly. This depends both on the arrangement of the MGDs within the MOSFET 100 and their characteristics. Typically, the losses during reverse mode decrease with increasing fraction of MGDs and are lower for a regular pattern arrangement of MGDs and MOSFETs compared to a clustered arrangement, i.e., an arrangement of MGDs and MOSFETs in different parts of the semiconductor device 100. A clustered arrangement of MOSFETS and MGDs may e.g., be used if the MOSFETS and MGDs have to be optimized differently.

[0107] Further, R_on will typically increase with increasing fraction of MGDs. The ratio between the MOSFETS and MGDs is typically chosen to be in a range between about 1:1 to 100:1 in regular pattern and clustered arrangements of MGDs and MOSFETS. Thereby, the trade-off between R_on and electric losses in reversed mode can be balanced in accordance with the MOSFET specifications for an application or circuitry.

[0108] In FIG. 17A the geometry and the current flow during reverse mode of a MGD is illustrated in more detail in the section 5 of FIG. 16. Typically, the p-type body region 50 includes a higher doped p-type contact portion 55. The thickness d₂ of the second insulating region 22 between the second gate electrode 21 at source potential and the body region 50, which is also connected to source, is for illustrative purposes higher than in FIGS. 14 and 15 and amounts to 35 nm. In FIG. 17B the hole current density (curve a), the electron current density (curve b) and the total current density (curve c) for the MGD of FIG. 17A are plotted as function of the voltage drop across the MGD. Due to the formed inversion channel within the body region 50, the total current is dominated by a unipolar electron current, i.e., the electron current contributes to more than 90% to the total current, above an average current flow density of about 10 μA/mm² in the common drift region 40. This depends on the thickness and/or the permitivitty of the second insulating region 22 between the second gate electrode 21 and the body region 50. In certain embodiments the current through the semiconductor device in reverse mode (forward bias of the body diode 15) is dominated by a unipolar current above an average current flow density in the drift region 40 of about 1 mA/mm².

[0109] FIG. 18 illustrates the current density-voltage characteristics of a silicon MGD as in FIG. 17A with poly-Si as material of the second gate electrode 21 and SiO₂ as material of the second insulating region 22 in dependence of the second thickness d₂. At given current density the voltage drop decreases with decreasing second thickness d₂. At a thickness d₂ of 5 nm, the current density-voltage characteristics of the MGD are, in a wide current density range, almost identical to a Trench-MOS-Barrier-Schottky-diode (TMBS-diode) with a work function of 4.75 eV. For a 3 nm thick gate oxide the losses are even lower. Thus an integrated MGD can replace an integrated Schottky-diode. Thereby, the aforementioned disadvantages of an integrated Schottky-diode can be avoided. Further, the losses in reverse mode can even further be reduced.

[0110] FIG. 19 illustrates the current density-voltage characteristics during reverse mode of an integrated power semiconductor device having a ratio between MOSFETS and MGDs of 9:1 in dependency of the second thickness d₂. The MOSFETS have a first thickness d₁ of 45 nm. Up to a current density of about 10 A/mm² the losses can be reduced significantly by the MGDs having a lower second thickness d₂.

[0111] In FIG. 20 the current per channel width—voltage characteristics of a typical silicon MGD as illustrated in FIG. 17A are plotted for second thicknesses d₂ of 5 nm, 8 nm and 35 nm. FIGS. 20A and 20B illustrate a linear plot and a linear-log plot, respectively. The threshold voltage V_th of a semiconductor device can be defined as the intersection of a suitable tangent with the abscissa in the linear plot as illustrated for the 8-nm-curve in FIG. 20A. This results in a threshold voltage V_th for the MGD having a second thickness d₂ of 8 nm of about 0.35 V. Another possibility of defining the threshold voltage V_th bases on a required current per channel width, which definition is used in this specification. For a current per channel width of 10 mA/m a threshold voltage V_th of about 0.26 V is obtained from FIG. 20B for the MGD with a second thickness d₂ of 8 nm. Accordingly, threshold voltages of MGDs can be achieved which are well below the values that can be obtained by using increasing the body effect for thicker gate oxides. In certain embodiments, the threshold voltage V_th of the MOS-gated diode, defined by a current per channel width of 10 mA/m, is positive but lower than or equal to about 0.26 V.

[0112] The threshold voltage V_th in FIG. 20 was obtained for SiO₂ having a relative dielectric constant of 3.9 as gate oxide, i.e., as material of the second insulating region 22. Hafnium oxide HfO₂ has e.g., a relative dielectric constant of about 12. Thus the curves illustrated in FIG. 20 also correspond to a MGD with HfO₂ as gate oxide but with an increased second thickness d₂ by a factor of about 3.1, which corresponds to the ratio between the relative dielectric constant of HfO₂ and 3.9. For example, the curves for the 8 nm thick SiO₂ gate oxide correspond also to the curves of a MGD which has a second gate electrode 21 insulated with an about 25.3 nm thick HfO₂ layer.

[0113] According to another embodiment, a semiconductor device includes a common source metallization, at least a first field-effect structure and at least a second field-effect structure. The first and second field-effect structure include a source region of a first conductivity type which is connected to the common source metallization and a body region of a second conductivity type which is adjacent to the source region. The first field-effect structure further includes a first gate electrode and a first insulating region of a first equivalent oxide thickness which is arranged at least between the first gate electrode and the body region. The second field-effect structure further includes a second gate electrode which is connected to the common source metallization and a second insulating region of a second equivalent oxide thickness which is arranged at least between the second gate electrode and the body region. The second equivalent oxide thickness is lower than the first equivalent oxide thickness.

[0114] In the context of the present specification, the term “equivalent oxide thickness” intends to describe the average thickness of the insulating region between a gate electrode and the body region multiplied with the ratio between the relative dielectric constant of the material of the insulating region and the relative dielectric constant of SiO₂ which is usually 3.9.

[0115] In certain embodiments the second equivalent oxide thickness is smaller than about 8 nm. In other words, the second gate capacitance per unit area C2 is larger than about 4.3 nF/mm² in certain embodiments. The second equivalent oxide thickness may also be smaller than 6 nm or 4 nm and may even be smaller than 1 nm. Likewise, the second gate capacitance per unit area C2 may be larger than about 5.7 nF/mm² or about 8.6 nF/mm² and may even be larger than about 34.4 nF/mm².

[0116] Integrated MGDs can also be used in reverse conducting IGBTs. FIG. 21 illustrates a similar semiconductor device as FIG. 3. However, instead of the common drain region 41 between the drift region 40 and the drain metallization 42 in FIG. 3 a highly doped p-type region 41a is arranged between the drift region 40 and the drain metallization 42 below the first trench 10. Thus four alternating layers (N—P—N—P along the dashed line 6) of an n-channel IGBT are formed. The additional PN junction blocks reverse current flow. This means that IGBTs cannot conduct in reverse mode, unlike a MOSFET. In bridge circuits, where a reverse current flow is needed, an additional diode (called a freewheeling diode) has to be connected antiparallel to the IGBT, i.e., in parallel to the body diode of the IGBT, to conduct a current in the opposite direction. Note that the source metallization 60 and the drain metallization 42 are also referred to as emitter metallization 60 and collector metallization 42 in case of an IGBT. Likewise, a highly doped n-type region 41b is arranged between the drift region 40 and the collector metallization 42 below the second trench 20. The second gate electrode 21, in contact with the emitter region 80 and the emitter metallization 60, the second insulating region 22, the body region 50 and the drift region 40, in contact with the collector metallization 42, form a MOS-gated diode. Typically, the MGD is connected in parallel to the body diodes 15 and can operate as an integrated freewheeling diode in reverse mode. Again, the capacitance per unit area between the second gate electrode 21 and the body region 50 is larger than the capacitance per unit area between the first gate electrode 21 and the body region 50. This can again be achieved by choosing the thickness d₂ to be smaller than the first thickness d₁ and/or by choosing a material with higher dielectric constant for the second insulating region 22 compared to the dielectric constant of the material of the first insulating region 12.

[0117] With reference to FIGS. 22-29, manufacturing processes according to several embodiments will be explained.

[0118] FIG. 22 illustrates a section of a silicon-semiconductor device 100 in a vertical cross-section after forming an n-type drain region 41 and after further processes including forming an n-type drift region 40, forming first and second trenches 10 and 20, forming insulating bottom portions 12c and 22c in the lower portions of the first and second trenches 10 and 20, respectively, forming respective field plates 16 and 26 and performing a thermal oxidation process to form a first dielectric SiO₂ region or layer 12a on the side walls in an upper portion of the first trenches 10. Typically, the dielectric layer 12a has a thickness of about 30 nm to about 60 nm and also covers the sidewalls of the second trenches 20. Thereafter, the first trenches 10 are covered with a photolithographically structured mask 7 to protect the first trenches 10. The resulting semiconductor structure is illustrated in FIG. 23.

[0119] According to FIG. 24, an optional ion implantation process, e.g., with P or As, can be carried out. Thereby, higher doped n-type doped regions 27 for reducing the channel length of the later formed second field-effect structure can be formed (see also FIG. 7). As illustrated in FIG. 25, a second optional ion implantation process, e.g., with P or As, can be carried out to form temporarily higher doped n-type regions 24. After a later ion implantation, e.g., with boron, and subsequent drive-in for forming the p-type body region, the higher doped n-type regions 24 are transformed into portions of the body region which have a lower effective p-type doping concentration than the other portions of the body region. Thereby the threshold voltage V_th of the later formed second field-effect structure can be reduced further.

[0120] Subsequently, the oxide on the side walls in the upper portion of the second trench is removed, e.g., by wet-chemical etching. Afterwards the mask 7 is removed. The resulting semiconductor structure is illustrated in FIG. 26. Thereafter, a second thermal oxidation process process is used to form a second insulating portion or dielectric layer 22a on the side walls in an upper portion of the second trench 20 as illustrated in FIG. 27. In the illustrated cross-section, the lateral thickness of the second dielectric layer 22a on the side walls in the upper portion of the second trench 20 ranges typically from about 1 nm to about 8 nm, but may be even smaller than 1 nm. Due to the different lateral thicknesses of the dielectric layers 12a and 22a on the side walls in the upper portions of the respective trenches, the later formed second field effect structure has a higher capacitance per unit area between its gate electrode and the body region 50 than the later formed first field-effect structure.

[0121] Thereafter, first and second gate electrodes 11 and 21 are formed, e.g., by a chemical vapor deposition (CVD) and back-etching of highly doped poly-Si. Further, a body region 50 and a source region 80 are formed, e.g., by appropriate ion implantation and subsequent drive-in. In addition, dielectric portions 70 are formed by deposition. Finally, a common gate metallization 65 in electrical contact with the first gate electrodes 11, a common drain metallization 42 in electrical contact with the drain region 41, and a common source metallization 60 in electrical contact with the body region 50, the source region 80, the second gate electrode 21 and the field plates 16 and 26 are formed. The resulting MOSFET with integrated MGDs 100 is illustrated in two different vertical cross-section in FIGS. 28 and 29, which correspond to the lines A and B of FIG. 30 respectively. The contact between the second gate electrodes 21 and the source metallization 60 is only illustrated in FIG. 29. FIGS. 30 and 31 illustrate plan views of the MOSFET 100 without and with the source metallization 60 and the gate metallization 65. The reference sign 220 denotes the portion of the semiconductor device 100 in which the second insulating portion or dielectric layer 22a was formed. In other words, the portion 220 of the semiconductor device 100 represents the portion in which the MGDs were formed. The reference signs 600 and 610 refer to the poly-Si filling, which forms the first and second gate electrodes 11 and 21, and to the groove contacts for connecting the body regions 50, the source regions 80 and the second gate electrode 21 with the source metallization 60, respectively.

[0122] In other words, the method described with reference to FIGS. 22-29 includes a process of providing a semiconductor body of a first conductivity type, e.g., of n-type. The semiconductor body typically includes a drift region 40 of the first conductivity type and first and second trenches 10 and 20. The first and second trenches 10 and 20 may already include suitably formed and insulated field plates 16 and 26, respectively. Further, the method includes forming a source region 80 of the first conductivity type and an adjoining body region 50 of a second, i.e., opposite, conductivity type. A first field-effect structure, which includes a first gate electrode 11 and a first insulating region 12a arranged at least between the first gate electrode 11 and the body region 50, and a second field-effect structure, which includes a second gate electrode 21 and a second insulating region 22a arranged at least between the second gate electrode 11 and the body region 50, are formed such that the capacitance per unit area between the second gate electrode 21 and the body region 50 is larger than the capacitance per unit area between the first gate electrode 11 and the body region 50. Further, a common source metallization 60 at least in contact to the source region 80 and the second gate electrode 21 is formed.

[0123] Typically, the formed second field-effect structure is a MGD connected in parallel to the body diode 15 of the first field-effect structure.

[0124] Examples for the first field-effect structure include but are not limited to a MOSFET and a reverse conducting IGBT. For forming an IGBT, the provided semiconductor body may already include highly doped region 41a of the second conductivity type and highly doped region 41b of the first conductivity type adjoining each other and arranged below the drift region 40.

[0125] In further embodiments, the provided semiconductor body already includes compensation structures as used in Superjunction-MOSFETs.

[0126] With reference to FIGS. 32-36, manufacturing processes according to certain embodiments will be explained. The FIGS. 32-35 illustrate vertical cross-sections through a semiconductor device 100 along the line B of FIG. 36. FIG. 32 illustrates the semiconductor device 100 after forming an n-type drift region 40 and after further processes including forming a p-type body region 50, forming an n-type source region 80, forming first and second trenches 10 and 20, forming insulating portions 70, forming shallow grooves 8 and forming higher doped p-type contact portions 55. The first and second trenches 10 and 20 include respective field plates 16 and 26, respective first and second gate electrodes 11 and 21 and respective first and second insulating regions 12 and 22. The higher doped contact portions 55 are arranged in the body region 50 below the adjoining grooves 8, and improve the later formed contact between the body region 50 and the source metallization 60. The trenches 10a and 10b are similar to the illustrated first trench 10. However, they are closest to a first lateral boundary of the semiconductor device 100 and have no adjoining source region 80 to compensate boundary effects. For the same reason, the trench 10b next to the first lateral boundary of the semiconductor device 100 and its field plate 16b extend vertically deeper into the drift region 40.

[0127] A photoresist 7 is deposited and structured such that only the grooves 8a next to the second trench 20 are exposed partially in a portion adjoining the second trench 20 and the contact portions 55 as illustrated in FIG. 33. The other grooves 8b remain completely filled with the photoresist 7. Subsequently, the insulating portion 70 which cover the second gate electrode 21 is removed by etching. This results in a structure 100 as illustrated in FIG. 34. Thereafter, the photoresist 7 is removed, and the source metallization 60 is deposited to electrically connect the source region 80, the body region 50 and the second gate electrode 21 as illustrated in FIG. 35. According to the performed manufacturing processes, the electrical connection between the source metallization 60 and the second gate electrode 21 is formed as a self-adjusted contact in a shallow trench 620. The shallow trench 620 extends vertically not as deep into the semiconductor device 100 as the grooves 8, which are also filled with the source metallization 60 to form a groove contact 610 to the body region 50, the source region 80 and the contact portion 55.

[0128] FIG. 36 illustrates a plan view of the MOSFET 100 including the region 220 in which the integrated MGD was formed. The reference sign 600 again refers to the poly-Si filling of the first and second trenches 10, 10a, 10b and 20. FIG. 36 further illustrates the shallow groove contacts 620 between the source metallization 60 and the second gate electrode 21. In addition, the groove contacts 610 between the source metallization 60 and the body region 50, the source region 80 and the contact portions 55 are illustrated.

[0129] FIGS. 37-73 illustrate vertical cross-sections through a semiconductor device 100 after several manufacturing processes for forming the semiconductor device 100.

[0130] The manufacturing processes illustrated with respect to FIGS. 37-48 avoid any lithographical processes on the gate oxide and allows the use of different materials, e.g., materials with different work functions, for the first and second gate electrodes 11 and 21. Starting point for the following process processes is the structure 100 of FIG. 22. Onto this structure a poly-Si layer 90 and a photoresist are deposited. Thereafter, the photoresist is structured for forming an etching mask 7. In the subsequent wet-chemical or dry poly-Si etching process the second trench 20 is exposed in an upper portion. The resulting semiconductor structure is illustrated in FIG. 37. Thereafter, the etching mask 7 is removed, the Silicon oxide is etched off the side walls of the second trench 20, and a thermal oxidation process is carried out to form the second insulating portions 22a. Afterwards, a second overlaying poly-Si layer 91 is deposited as illustrated in FIG. 38. Subsequently, a chemical-mechanical polishing (CMP) process is carried out to remove the poly-Si above the first and second trenches 10 and 20 and to form a flat surface.

[0131] In an alternative process, a poly-Si layer 90 is deposited onto the structure illustrated in FIG. 22 and subsequently etched back. On top of the surface of the resulting structure a Si₃N₄ layer 92 is deposited, e.g., by a CVD process. This results in a structure as illustrated in FIG. 39. In the next two processes the body region 50 and source region 80 are formed by appropriate ion implantation and drive-in. This results in a structure as illustrated in FIG. 40. Thereafter, the following processes are subsequently performed. The Si₃N₄ layer 92 is removed by a wet-chemical etching. The first trenches 10 are masked by a further photolithographically structured mask 7b. The poly-Si 91 is removed from the second trench 20 by etching. Further, the insulating oxide layer on the sidewalls of the second trench 20 and on the source region 80 adjoining the second trench 20 are removed using an isotropic etching process. The resulting structure is illustrated in FIG. 41. After removing the mask 7b, a thermal oxidation is carried out to form second insulating portion 22a on the side walls in the upper portion of the second trench 20. Typically, the second insulating portions 22a are, in the lateral direction of the vertical cross-section illustrated in FIG. 42, thinner than the first insulating portions 12a. Subsequently, the second gate electrode 21 is formed by CVD and back-etching of highly doped poly-Si or a material with a lower work function than highly doped poly-Si such as TiN. This results in the structure illustrated in FIG. 43.

[0132] Finally, dielectric portions 70, a gate metallization 65 in contact with the first gate electrodes 11, a drain metallization 42 in contact with the drain region 41, and a source metallization 60 in contact with the second gate electrode 21, the body region 50 and the source region 80 are formed.

[0133] Alternatively, the following processes can be carried out after the processes resulting in the structure illustrated in FIG. 40. On top of the Si₃N₄ layer 92 an intermediate oxide layer 93 is deposited and a photolithographically structured mask 7b is formed thereon for masking the first trench 10 as illustrated in FIG. 44. Subsequent etching of the intermediate oxide layer 93, removing of the mask 7b and wet chemical etching of the Si₃N₄ layer 92 selective to SiO₂ result in a structure 100 as illustrated in FIG. 45. Thereafter, the poly-Si 91 is removed in the upper portion of the second trench 20 using isotropic etching. Further, the insulating silicon oxide layer on the sidewalls of the second trench 20 and on the source region 80 adjoining the second trench 20 is removed by isotropic etching. The resulting structure is illustrated in FIG. 46. Subsequently, a thermal oxidation is carried out to form second insulating portions 22a on the side walls in the upper portion of the second trench 20. The second insulating portions 22a are, in the lateral direction of the vertical cross-section illustrated in FIG. 47, typically thinner than the first insulating portion 12a.

[0134] Subsequently, the SiO₂ layer formed during the thermal oxidation on the source region 80 is anisotropically etched back to expose the upper surface of the source region 80 for later contacting to the source metallization 60. Alternatively, an isotropic etching in combination with a protecting plug can be used.

[0135] Finally, the connected second gate electrode 21 and source metallization 60 are formed by depositing of highly doped poly-Si or a material with a lower work function than highly doped poly-Si such as TiN. The resulting MOSFET with integrated MGDs is illustrated in FIG. 48. The illustrated dashed line indicates that the second gate electrode 21 and the gate metallization 60 can be made of the same material or different materials.

[0136] With respect to FIGS. 49-73 five embodiments of manufacturing methods for forming a field plate trench semiconductor device 100 will be explained. They all have in common that at least the sidewalls of the second trenches 20 are protected by a protecting region against oxidation during the thermal oxidation for forming the first insulating regions 12a in an upper portion of the first trenches 10. Otherwise, the thermal oxidation for forming the first insulating regions 12a will also cause oxidation of the silicon in the boundary region to the second trench 20. Typically, this causes a formation of processes in the mesa next to already formed field plates. Depending on the size and the position of the formed mesa processes, high field strength may occur in particular during reversed bias. Therefore, a material of low oxygen permeability under thermal oxidation conditions is typically deposited at least on the side walls in an upper portion of the second trench 20. Thereby, the formation of a process in the drift region 40 next to a transition region between the second insulating portion 22a and the thicker insulating bottom portion 22c of the second insulating region 22 can be avoided or at least reduced to a size which is smaller than about the half of the thickness d₂ of the second insulating portion 22a between the second gate electrode 21 and the body region 50. For example, the size of the process in the mesa may be only 4 nm or 2 nm or even smaller. Thus the size of the process in the mesa is significantly reduced compared to standard processing of MOSFETS which results in mesa processes of about 20 nm. Thereby, the electric field magnitude during reverse bias can be reduced in the drift region 40 close to the transition region. This will be explained with respect to FIGS. 74A-F.

[0137] In short, the embodiments include providing a semiconductor substrate of a first conductivity type. At least a first trench 10 and at least a second trench 20 are etched into the semiconductor substrate. A first oxide layer which covers at least a lower portion of the walls of the first trench 10 and a lower portion of the walls of the second trench 20 is formed. Subsequently, a first conductive region 16 at least in the lower portion of the first trench 10 and at least a second conductive region 26 in the lower portion of the second trench 20 are formed. This is typically done by CVD and back-etching of highly conductive poly-Si. A thermal oxidation process is performed to form a first insulating region 12a on the side walls in an upper portion of the first trench 10. During this thermal oxidation process the second trench 20 is protected such that the semiconductor substrate forming the walls of the second trench 20 is not or almost not oxidized. Thereafter, a second insulating region 22a is formed on the side walls in an upper portion of the second trench 20. Subsequently, a first gate electrode 11 in the upper portion of the first trench 10 and a second gate electrode 21 in the upper portion of the second trench 20 are formed. Further, a source region 80 of the first conductivity type and a body region 50 of a second conductivity type are formed such that they are adjoining. Typically, the drift region 40 adjoining the body region 50 is thereby finally formed such that the first and second trenches 10 and 20 extend in a vertical direction below the pn-junction between the body region 50 and the drift region 40. Thereafter, a source metallization 60, which is at least in contact to the source region 80 and the second gate electrode 21, is formed. Typically, the body region 50 is also electrically connected to the source metallization 60. Further, the first and second conductive region 16 and 26 are typically also electrically connected to the source metallization 60 and operate as field plates 16 and 26.

[0138] According to an embodiment the first insulating region 12a and second insulating region 22a are formed such, that the capacitance per unit area between the second gate electrode 21 and the body region 50 is higher than the capacitance per unit area between the first gate electrode 11 and the body region 50.

[0139] According to a further embodiment, the second gate electrode 21 extends less deep into the drift region 40 than the first gate electrode 11. This further reduces the electric field magnitude during reverse mode in the drift region 40 next to the transition region of the second insulating region 22. Typically, this is also achieved by the five methods for forming a field plate trench semiconductor device explained in the following.

[0140] Referring now to FIGS. 49-56 the first of the five embodiments for forming a field plate trench semiconductor device 100 will be explained in detail. FIG. 49 illustrates the structure 100 after providing an n-type Si-substrate which includes a drift region 40 and a higher doped drain region 41 and after further processes including etching of two first trenches 10 and a second trench 20 into the semiconductor substrate, forming a first oxide layer 71 on the semiconductor substrate such that the side walls and the bottom wall of the first and second trench 10 and 20 are also covered and forming first conductive regions 16 in a lower portion of the first trench 10 and a second conductive region 26 in a lower portion of the second trench 20. Typically, first and second insulating bottom portion 12c and 22c are later at least partially formed from the lower portions of the first oxide layer 71 next to the first and second conductive regions 16 and 26, respectively. Further, the first and second conductive region 16 and 26 are typically formed by CVD and back-etching of highly doped poly-Si. Thereafter, a Si₃N₄ mask 92 is formed by providing a nitride layer overlying the first oxide layer 71, using e.g., CVD and subsequent structuring of the nitride layer. The resulting structure 100 is illustrated in FIG. 50. The Si₃N₄ mask 92 is used to protect the second trench 20 against the subsequent etching of the first oxide layer 71. This etching process results in exposing the upper surface of the first conductive regions 16 and the side walls of the first trenches 10 in an upper portion as illustrated in FIG. 51. This also leads to the formation of first insulating bottom portion 12c in the lower portion of the first trenches 10. Now a first insulating portion 12a can be formed by thermal oxidation on the side walls in the upper portion of the first trench 10. As can be seen in FIG. 52 the thermal oxidation process results in the formation of processes in the drift region 40 next to the transition regions 13 between the first insulating bottom portion 12c and the first insulating portion 12a of the first insulating region. In other words, the vertical boundaries between the mesa and the first insulating region 12, which is formed by the insulating bottom portion 12c and the first insulating portion 12a, deviates from the illustrated straight lines f. In the illustrated cross-section, the lateral process of the mesa next to the transition regions 13 is typically about the half of the lateral thickness of the first insulating portion 12a. The sidewalls of the second trench 20 are, however, completely protected by the remaining portions of the first oxide layer 71 and the mask 92 against thermal oxidation. Thus, during the process of performing the thermal oxidation for forming the first insulating portion 12a on the side walls in the upper portion of the first trench 10, the Si-substrate forming the side walls of the second trench 20 is practically not oxidized. Thereby, the formation of a process in the mesa close to second trench 20 can be avoided.

[0141] During thermal oxidation, third insulating portions 12b are typically also formed on the first conductive regions 16. Typically, the third insulating portions 12b are thicker than the first insulating portions 12a in the direction of the oxide growth. This is because the growth rate of the thermal oxide is higher for highly doped poly-Si used as material of the conductive regions 16 compared to the weaker doped silicon of the drift region 40.

[0142] Thereafter, the first trenches 10 are masked and subsequently an etching process is performed to expose the upper portion of the second trench 20. FIG. 53 illustrates in addition to FIG. 52 a photolithographically structured mask 7, which protects the first trenches 10. FIG. 54 illustrates the structure 100 after a HF-dipping and an isotropic plasma etching to remove the Si₃N₄ mask 92 and the exposed oxide. Thereby, the upper surface of the second conductive regions 26 and the side walls of the second trenches 20 are exposed in the upper portions the second trenches 20. Thereafter, the mask 7 is removed and a second insulating region 22a is formed on the side walls in the upper portion of the second trench 20 by thermal oxidation as illustrated in FIG. 55. Alternatively and/or in addition, a dielectric material such as Si₃N₄, SiO_xN_y or HfO₂ may be deposited on the side walls in the upper portion of the second trench 20 to form the insulating portion 22a.

[0143] Typically, the second insulating portion 22a is formed such, that it has a higher dielectric constant and/or lower thickness in the lateral direction of the cross-section illustrated in FIG. 55 than the first insulating portions 12a.

[0144] In addition, a fourth insulating portion 22b is typically formed during formation the second insulating portions 22a. The thickness of the fourth insulating portions 22b in the vertical direction may be adjusted e.g., by deposition of further dielectric material such that a later formed second gate electrodes 21 extends e.g., 50 nm or 100 nm less deep into the drift region 40 than a later formed first gate electrode 11.

[0145] Typically, the upper surface of the mesas adjoining the second trench 20 is also covered with a dielectric layer which was also formed during the process of forming the insulating portions 22a.

[0146] Subsequently, highly doped poly-Si is deposited and etched back to fill the first and second trenches 10 and 20 at least partially. Thereby, first and second gate electrodes 11 and 21 are formed in the upper part of the first trench 10 and second trench 20, respectively. The resulting semiconductor device structure 100 is illustrated in FIG. 56.

[0147] Typically, the minimum distance between the first field plate 16 and the first gate electrode 11 is larger than the minimum distance between the second field plate 26 and the second gate electrode 21.

[0148] Thereafter, a source region 80 and an adjoining body region 50 are formed by ion implanting. A source metallization 60 at least in contact to the source region 80 and the second gate electrode 21 is formed using standard techniques. At least a part of the typically formed dielectric layer on top of the mesas adjoining the second trench 20 is typically removed, e.g., by etching, prior to depositing the source metallization 60.

[0149] The second embodiment for forming a field plate trench semiconductor device 100 includes the same initial process processes which result in the semiconductor structure illustrated in FIG. 52. Thereafter, a HF-dipping and an isotropic nitride etching e.g., with hot phosphoric acid is performed to remove the Si₃N₄ mask 92 as illustrated in FIG. 57. A photolithographically structured mask 7b which protects the first trenches 10 is formed on the semiconductor structure 100. Subsequently, the oxide layer is etched back to expose an upper portion of the second trench 20. The resulting structure 100 is illustrated in FIG. 58. Thereafter, the mask 7b is removed and a typically thin second insulating region 22a is formed on the side walls in the upper portion of the second trench 20 by a thermal oxidation process or by a CVD process. Thus a similar semiconductor structure as already illustrated in FIG. 55 is obtained. Again, the second insulating region 22a may have a higher effective permittivity than the first insulating region 12a.

[0150] The subsequent manufacturing processes for forming a field plate trench semiconductor device are in accordance with the manufacturing processes which have been explained above with respect to FIG. 56.

[0151] With respect to FIGS. 59 to 63 a further embodiment for forming a field plate trench semiconductor device will be explained. FIG. 59 illustrates the structure 100 after providing an n-type Si-substrate with a drift region 40 and a higher doped drain region 41 and after further processes including etching of at least a first trenches 10 and at least a second trench 20 into the semiconductor substrate, forming a first oxide layer 71 on the semiconductor substrate such that the walls of the first and second trench 10 and 20 are also covered and forming first conductive region 16 in a lower portion of the first trench 10 and a second conductive region 26 in a lower portion of the second trench 20. Typically, first and second conductive regions 16 and 26 are formed by the following subsequent processes: CVD of highly doped poly-Si, masking the second trench 20 with a first photolithographically structured mask 7, and back-etching of highly doped poly-Si. Thereafter, the first photolithographically structured mask 7 used for protecting the second trench 20 during the back etching of the poly-Si is removed. A second photolithographically structured mask 7a is formed to protect the second trench 20 during the subsequent etching of the first oxide layer 71. The resulting structure 100 is illustrated in FIG. 60. The second photolithographically structured mask 7a is removed and a thermal oxidation process is performed to form insulating portions 12a on the side walls of the upper part of the first trenches 10. During thermal oxidation process the sidewalls of the second trench 20 are completely protected against thermal oxidation by the remaining portions of the first oxide layer 71 and the highly doped poly-Si in the second trench 20. In parallel, third insulating portions 12b on the upper surface of the first conductive regions 16 are formed and the first insulating region 71 is closed by a formed insulating portion 71b on the second conductive region 26 as indicated in FIG. 61. In other words, the second trench 20 is completely filled with the first oxide layer 71 and the second conductive region 26 during the thermal oxidation process for forming the first and third insulating portions 12a and 12b of the first insulating region 12.

[0152] FIG. 61 further illustrates a photolithographically structured mask 7b which protects the first trenches 10 during the subsequent process of disposing an upper portion of the second trench 20 by etching the oxide layer. Thereafter, the mask 7b is removed, and second insulating portions 22a on the side walls of the upper portion of the second trench 20 are formed by a further thermal oxidation process and/or by CVD. In parallel, fourth insulating portions 22b are typically formed on the exposed surfaces of the second conductive regions 26 as indicated in FIG. 62. Further, the fourth insulating portions 22b may also be formed on the second insulating bottom portion 22c, e.g., by CVD. Thereafter, first and second gate electrodes 11 and 21 are formed by deposition of e.g., highly doped poly-Si and subsequent back etching of the deposited poly-Si. In the cross-section of the structure 100 illustrated in FIG. 63, the second gate electrode is not simply connected. Further, the first gate electrodes 11 extends deeper into the drift region 40 than the second gate electrodes 21. Typically, the first gate electrodes 11 extend more than 50 μm, e.g., 100 μm, deeper into the drift region 40 than the second gate electrodes 21. The subsequent manufacturing processes for forming a field plate trench semiconductor device are again similar to those explained with respect to FIG. 56.

[0153] The fourth embodiment for forming a field plate trench semiconductor device is similar to the previous method up to the processes resulting in the structure 100 of FIG. 61. However, the mask 7b, which protects the first trenches 10, has, in the illustrated cross-section of FIG. 64, a smaller opening above the second trench 20. In FIG. 64 the opening only exposes the insulating portion 71b of the first oxide layer 71 above the second conductive region 26. Subsequently, an oxide etching process is performed to remove the insulating portion 71b as illustrated in FIG. 65. Thereafter, the poly-Si in the second trench 20 is etched back followed by a further oxide etching process to expose an upper part of the second trench 20. This is illustrated in FIG. 66. Subsequently, the photolithographically structured mask 7b is removed. A further thermal oxidation process and/or a further CVD process is performed to form second insulating portions 22a on the side walls of the upper part of the second trench 20 and fourth insulating portions 22b on an upper surface of the second conductive regions 26. This results in a semiconductor structure as illustrated in FIG. 67. Thereafter, first and second gate electrodes 11 and 21 are formed by deposition of e.g., highly doped poly-Si and subsequent back etching of the deposited highly doped poly-Si. This is illustrated in FIG. 68. The subsequent manufacturing processes for forming a field plate trench semiconductor device have already explained with respect to FIG. 56.

[0154] The fifth embodiment for forming a field plate trench semiconductor device 100 include the same initial process processes resulting in the semiconductor structure 100 illustrated in FIG. 59. Thereafter, the first mask 7 is removed and the first oxide layer 71 is etched back. This results in the semiconductor structure 100 of FIG. 69. A thermal oxidation is carried out for forming first insulating portions 12a on the side walls in the upper portion of the first trenches 10 as illustrated in FIG. 70. During the thermal oxidation process the sidewalls of the second trench 20 are completely protected against thermal oxidation by the remaining portions of the first oxide layer 71 and the highly doped poly-Si in the second trench 20. In parallel, an oxide layer 12d is formed on an upper surface of the mesas and the second conductive region 26. A photolithographically structured mask 7b is formed, which protects the first trenches 10 during the subsequent process of disposing an upper portion of the second trench 20 by etching the oxide layer 12d and back-etching of the first oxide layer 71. After subsequent removing of mask 7b, second insulating portions 22a are formed by a further thermal oxidation process and/or by CVD on the side walls of the upper portion of the second trench 20. In parallel, fourth insulating portions 22b are typically formed on the upper surface of the second conductive regions 26 and on the second insulating bottom portion 22c as indicated in FIG. 72. Thereafter, first and second gate electrodes 11 and 21 are formed by deposition of e.g., highly doped poly-Si and subsequent back-etching of the deposited highly doped poly-Si as illustrated in FIG. 73. The subsequent manufacturing processes for forming a field plate trench semiconductor device are similar to those explained with respect to FIG. 56.

[0155] Referring now to FIG. 74, the improved performance of the semiconductor devices 100, which are produced in accordance with the above described embodiments for forming a field plate trench semiconductor device, will be explained. During reverse mode and higher load, the semiconductor devices 100 may be driven into an Avalanche mode. An Avalanche process during reverse mode may result in entrapment of charges in the gate oxide or gate insulation. This is likely to change the characteristics such as forward voltage drop of the semiconductor device in forward mode. Therefore, it is desirable to avoid high field strength close to a thin gate oxide. FIGS. 74A-D illustrate the magnitude of the electric field as an overlay of a contour plot and a density plot (darker regions correspond to higher electric field magnitudes) with linear scaling in reverse mode. Vertical cross-sections through four different devices are compared. Each of the sections 5a includes a mesa and the half of the respective adjoining trenches. For the simulation, a vanishing current was assumed at the illustrated lateral boundaries. At the lower and the upper vertical boundary the electric potential is fixed to drain potential and source potential, respectively. The voltage difference between drain and source was V_DS=30V. Further, the upper boundary of the sections 5a goes through the interface between the body region 50 and the source region 80. FIG. 74A illustrates the electric field magnitude between two equal MOSFET field-effect structures having a 35 nm thick SiO₂ gate oxide between the body region 50 and the respective gate electrode 11. This structure is also referred to as single gate oxide structure. Further, two lines e and f are drawn. They cross close to the left process of the mesa adjoining the transition region of the first insulating region 12 from a first insulating portion between the gate electrode 11 and the body region 50 to a first insulating bottom portion between a first field plate 16 and the drift region 40. The first insulating bottom portion has a larger lateral extension than the first insulating portion. In other words, the transition region of the first insulating region 12 is the region next to the first gate electrode 11, in which the lateral extension, in the illustrated cross-section, of the first insulating region 12 changes. The transition regions are typically close to the transition between an essentially vertical boundary between the gate electrode 11 and the gate oxide 12 and an essentially horizontal or lateral lower boundary between the gate electrode 11 and the gate oxide 12. For clarity reasons, only one of the two processes in FIG. 74A is designated with the reference sign 9. In FIG. 74B the electric field magnitude is plotted for the mesa between a MGD on the left and a MOSFET on the right. The MGD has a 5 nm thick gate oxide between the body region 50 and the second gate electrode 21 and the MOSFET has a 35 nm thick gate oxide between the body region 50 and the first gate electrode 11. The two lines e and f cross close to the left process of the mesa adjoining the illustrated right transition region of the second insulating region 22 from the 5 nm thick portion between the gate electrode 12 and the body region 50 to a lower thicker portion. This structure is denoted in FIG. 74 as dual gate oxide structure. The structure illustrated in FIG. 74C is similar to the one illustrated in FIG. 74B, but the left process of the mesa adjoining the right transition region was avoided during manufacturing as has been explained above. Therefore, this structure is denoted in FIG. 74 as dual gate oxide no process structure. In other word, the mesa is, in the illustrated cross-section, practically straight down to a vertical depth into which the field plates 16 and 26 extend to. The structure illustrated in FIG. 74D is similar to the one illustrated in FIG. 74C, i.e., the left process of the mesa adjoining the right transition region was avoided during manufacturing. Further, the second gate electrode 21 extends not as deep into the drift region 40 as the first gate electrode 11. The first gate electrode 11 extends about 100 μm deeper into the drift region 40 compared to the second gate electrode 21. Consequently, the right transition region of the second insulating region 22 is also arranged higher by a vertical distance dy of about 100 μm. This structure is denoted in FIG. 74 as dual gate oxide no process II structure. FIG. 74E illustrates the field magnitude along the lines e from top to bottom of the lines e in the mesa. FIG. 74F illustrates the field magnitude along lines f from top left to right bottom of the lines e in the mesa. In both FIGS. 74E and 74F the curves a, b, c and d correspond to the FIGS. 74A, 74B, 74C and 74D, respectively. As can be appreciated from the heights of the first peaks of the curves illustrated in FIG. 74E and the curves in FIG. 74F the electric field magnitude in the transition region close to the thin gate oxide of the MGD can be reduced by avoiding a process in the mesa and/or by extending the first gate electrode 11 deeper into the drift region 40 than the second gate electrode 21. Thereby, the charge generation, the risk of charge entrapment in the gate oxide and the risk of latch-up of the MGD during reversed current flow and Avalanche conditions can be reduced.

[0156] Although specific embodiments have been illustrated and described herein, it will be appreciated by those of ordinary skill in the art that a variety of alternate and/or equivalent implementations may be substituted for the specific embodiments shown and described without departing from the scope of the present invention. This application is intended to cover any adaptations or variations of the specific embodiments discussed herein. Therefore, it is intended that this invention be limited only by the claims and the equivalents thereof.