WO2014177612A1

WO2014177612A1 - Method for manufacturing a semiconductor device comprising a thin semiconductor wafer

Info

Publication number: WO2014177612A1
Application number: PCT/EP2014/058820
Authority: WO
Inventors: Wolfgang Janisch
Original assignee: Abb Technology Ag
Priority date: 2013-04-30
Filing date: 2014-04-30
Publication date: 2014-11-06

Abstract

The present invention is concerned with the field of power electronics, such as a method for producing a semiconductor device comprising a thin semiconductor wafer 1 as a device element. The invention describes a method for manufacturing a semiconductor device comprising a thin semiconductor wafer 1, wherein the thin wafer contains the device element in the finalized device, e.g. active layers, buffer layers etc., and the method is comprised of an initial thinning of a first wafer 2, which forms the thin semiconductor wafer 1, to a final thickness of the thin semiconductor wafer 1. In a next step the semiconductor wafer 1 is permanently bonded to a carrier wafer 2 or element. This way the semiconductor wafer 1 and the carrier wafer 3 form a stack which can be easily handled in further topology or device production processing such as front-end-of line processing. In a final step or in a finalizing step of the device production the carrier wafer 3 is destructively removed.

Description

DESCRIPTION

Method for Manufacturing a Semiconductor Device Comprising a Thin

Semiconductor wafer

FIELD OF THE INVENTION

The invention relates to the field of power electronics, such as a method for producing a semiconductor device comprising a thin semiconductor wafer as a device element.

BACKGROUND OF THE INVENTION

Manufacturing semiconductor devices is evolving as production efficiency is to be increased while production costs have to be lowered at the same time. As part of this development production for instance of insulated-gate bipolar transistors IGBTs or diodes are transferred to large wafers such as 200 mm or even 300 mm diameter wafers. In the production process handling wafers with large diameter is difficult as these processes involve applying pressure to the wafer and frequent transferring of the wafer. In particular for devices with thin films, e.g. as a device layer or element, the handling of these thin or thinned wafers may lead easily to breaking of the wafer and increase costs.

To avoid breakage of the wafer, thinning of the wafer is according to the state of the art in a final section of the production process, in particular after topology is formed. Thinning usually includes a combination of mechanical grinding and chemical etching. The grinding process applies pressure on the stack of layers forming the device and leaves an imprint of any topology or structure formed on the surface or in layers of the stack. This way a thickness of the thin layer cannot be controlled in order to achieve high quality and reproducible devices as the thickness varies with the structure of the thin layer.

In addition, for a 150 mm wafer the buffer layer on an anode side of an IGBT may be formed by a more than 200 μιη deep diffusion on the initial material. However, for a 200 mm or a 300 mm wafer deep diffusion is not feasible anymore.

Further an epitaxial layer may be used for creating a thin semiconductor wafer device containing a soft deep buffer, e.g. for a 1200 V device a 100 pm and for 1700 V device a 150 Mm layer has to be grown. However, using epitaxy is very cost intensive and, thus generally not suited for lowering the production cost of for instance IGBTs or diodes. DESCRIPTION OF THE INVENTION

It is therefore an objective of the invention to provide a method for production of high quality devices at low costs. This objective is achieved by a method according to claim 1 . Preferred embodiments are evident from the dependent patent claims, wherein the claim dependency shall not be construed as excluding further meaningful claim combinations.

According to the invention, a method for manufacturing a semiconductor device, exemplary an insulated gate bipolar transistor (IGBT) or a diode, comprising a semiconductor wafer, wherein in the finalized device the thin semiconductor wafer contains the device element, e.g. active layers, buffer layers etc., is provided. The method is comprised of an initial thinning of a first wafer or device wafer to a final thickness, thereby forming the semiconductor wafer. In a next step the semiconductor wafer is bonded to a carrier wafer or carrier element by a permanent bonding, wherein exemplarily a first oxide layer is arranged on the bonding side of the first wafer or the carrier element. This way the first wafer and the carrier wafer form a stack which can be easily handled in further topology or device production processing such as front-end-of line processing. In a final (later) section or in a finalizing step of device production the carrier wafer is destructively removed.

A permanent bonding is needed in this step for achieving a reliable connection between the semiconductor wafer and the carrier wafer, which can withstand the high temperatures required for the creation of the front sided layer in power semiconductor devices. In an exemplary embodiment, temperatures of at least 900 °C are applied for a diffusion step for the creation of a layer of another conductivity type than the wafer (i.e. the drift layer, which is such part of the semiconductor wafer of unamended doping concentration in the finalized power semiconductor device). The carrier wafer is removed after topology production processing on the first side destructively to separate the semiconductor wafer from the carrier wafer.

In an exemplary variant of the invention, the first wafer is a Float Zone FZ wafer and/or the carrier element is a Czochralski wafer. A FZ wafer is produced using a float zone process and the Czochralski wafer is produced using a Czochralski process.

In a further variant of the invention, the first wafer is a non-prediffused wafer, i.e. there is basically no topology on the wafer. This way a very homogeneous final thickness is achieved. A non-pre-diffused wafer shall have a substantially constant and homogeneous doping concentration, which means that the doping concentration is substantially homogeneous throughout the wafer, however without excluding that fluctuations in the doping concentration within the wafer being in the order of a factor of one to five may be possibly present due to e.g. a manufacturing process of the wafer being used. The part of the semiconductor wafer having unamended doping concentration in the finalized device may form a drift layer, also called (n-) base or drift layer. An exemplary doping concentration of the first wafer 2/semiconductor wafer/drift layer is between 5 ^* 10¹² cm ³ and 1 .5 ^* 10¹⁴ cm ³.

In an exemplary embodiment of the invention, the method comprises further partial or all area applying ions, i.e. implanting, for instance of n-type and/or p-type implants to form a buffer, cathode and/or an anode (for creation of an IGBT) on the first wafer before bonding of the semiconductor wafer to the carrier element, wherein subsequent annealing and/or driving of the implant may be carried out. Applying ions partial or over the whole area (all area) can be understood that ions are applied over a whole side (or partially, e.g. by masking) of the semiconductor wafer. The side is exemplarily the side, on which the semiconductor wafer is to be bonded to the carrier element or the side opposite to the bonding side.

In yet another embodiment of the invention applying ions forms a double buffer structure, for instance carried out by two separate applications and diffusion of the same ions, wherein the diffusion can be performed partially, but not completely together, or separately or simultaneously applying two different ions, e.g. ions with different diffusion rates, and diffusing these ions either partially or completely separately (for separate application of ions) or diffusing the different ions simultaneously (for different ions either separately or simultaneously applied).

In a further variant of the invention the bonding is between an oxide layer, e.g. greater than 0.1 pm and lower than 3 μιη, on the bonding side of the carrier element and the first wafer. Further, the oxide layer servers as a solid barrier for impurities of the carrier wafer during topology production processing. The oxide layer may be formed on the first wafer/semiconductor wafer and/or on the carrier element.

In another embodiment of the invention, the method comprises further producing an implanted carrier wafer by partial or all area implanting of the carrier element. In a further variant, the implanted carrier element serves as a dopant source for the backside of semiconductor wafer.

In a further preferred embodiment, removing of the carrier element is stopped at or nearby the bonding interface by an electrical control end point. This may be achieved by detecting, e.g. electrically, the oxide layer at the bonding interface.

In a further preferred embodiment, the method comprises further front-end of line FEOL processing, e.g. fabrication of individual devices, and back-end of line BEOL processing, e.g. IGBT anode implanting, metal stacking, sintering, and annealing.

In yet another embodiment, the semiconductor device is a bipolar transistor, in particular an insulated-gate bipolar transistor IGBT, or a diode. In another embodiment, the thinning of the first wafer and removing of the carrier wafer is achieved by grinding, e.g. mechanical grinding and/or etching, e.g. chemical etching.

This way the invention allows for a low cost production using well established technology for producing semiconductor devices on Float Zone wafers and guarantees controlled thickness for minimizing wafer to wafer variation. In addition, the stack may be formed by using relatively cheap wafers such as Czochralski wafers as carrier wafers.

BRIEF DESCRIPTION OF THE DRAWINGS

The subject matter of the invention will be explained in more detail in the following text with reference to preferred exemplary embodiments which are illustrated in the attached drawings, in which:

Fig. 1 schematically shows the thinning of a first wafer;

Fig. 2 schematically shows applying of ions on the first wafer;

Fig. 3 shows a carrier wafer bonded to the first wafer; and

Fig. 4 shows final steps of device production.

The reference symbols used in the drawings, and their meanings, are listed in summary form in the list of designations. In principle, identical parts are provided with the same reference symbols in the figures. DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

Fig.1 shows schematically an initial step of thinning for manufacturing a semiconductor device comprising a thin (n-) doped semiconductor wafer 1 having a first side and a second side opposite to the first side, wherein a first wafer 2 is thinned to a desired thickness of the semiconductor wafer 1 , e.g. by mechanical grinding, chemical etching such as SEZ-etching, or a combination of these. For instance the thickness of the semiconductor wafer 1 may be from 1 15 Mm for a device suitable up to 1 .2 kV or up to 380 μιη for a device suitable up to 3.3 kV, exemplarily the semiconductor wafer has a thickness between 100 to 600 μιη. By thinning in the beginning of the production process, the thickness of the semiconductor wafer may be controlled for instance up to an accuracy of ± 1 μιη. Exemplarily, the wafer 1 is thinned on the first and second side of the wafer. In particular, as the first wafer 2 has no topology on its surface the thickness is highly accurate. A polishing step may follow the thinning step. Exemplarily, the wafer is thinned (and polished) on one side and only polished on the other side. After the initial thinning to the final wafer thickness further processing such as n buffer implantation may be carried out on one of the sides, exemplarily on a thinned side (backside or second side opposite to a front side, i.e. a first side) of the first wafer 2 and/or p implantation (for the creation of a p doped anode layer in case of the manufacturing of an IGBT). Then, a carrier wafer 3 may be bonded on the second side of the wafer (e.g. the thinned surface of the first wafer 2) as a permanent bonding. Only a permanent bonding makes it possible to perform high temperature treatment steps in the following manufacturing of layers on the first side of the wafer. The permanent bonding may be made by pressing the wafer 1 to the carrier wafer 3 at high pressure. Afterwards or simultaneously a bond anneal step may be performed at a temperature of at least 300 °C. Exemplarily, temperatures of at least 700 °C are applied. A maximum temperature at this step may exemplarily be 800 °C.

The term implantation in this document shall be understood as any way to apply ions to the wafer and shall also cover deposition of ions.

In addition, the first wafer 2 or semiconductor wafer 1 may be protected by an additional oxide layer 4 on the front side (first side) not to be bonded to the carrier wafer 3 which may be greater, i.e. thicker, than 0.1 μιη and lower, i.e. thinner, than 3 μιη. The back side of the semiconductor wafer 1 or the side of the carrier wafer 3 to be bonded to the semiconductor wafer 1 bonded to the carrier wafer 3 may carry a further oxide layer, not shown, which should be lower, i.e. thinner, than 0.1 μιη. The arrow in Fig. 1 indicates the side of processing and in particular the thinning of the first wafer 2.

Fig. 2 shows applying of ions, i.e. implanting, to the semiconductor wafer 1 before bonding the carrier wafer 3 on the wafer 1 (i.e. after thinning of the first wafer 2). The arrow in Fig. 2 indicates the side of processing and in particular the applying of ions on the first wafer 2. Applying ions is used to form a buffer, e.g. a deep diffused buffer. In addition, a double buffer may be formed by using subsequent buffer implanting with an annealing step in-between applying ions or the double buffer may be formed by applying simultaneously ions with different diffusion rates. In addition, the buffer softness may be tuned and an additional drive for very deep diffused buffer or for low voltage application of ions may be carried out.

Fig. 3 shows the carrier wafer 3 bonded to the wafer 1 . In one embodiment of the invention, an oxide layer on the bonding side of the carrier wafer 3 and/or first wafer 2, not shown, may serve further as a solid barrier for impurities of the carrier wafer 3 during topology production processing. In a further embodiment, the carrier wafer 3 may be partially or all area implanted, where the carrier wafer 3 then may serve as a dopant source for the wafer 1 to build up an accurately defined deep anode in the buffer area of wafer 1 (for an IGBT) and/or a buffer layer. In the oxide layer, a dopant may be added for the creation of a buffer layer and/or an anode layer (for creation of an IGBT). It is also possible to combine steps like creating a first buffer layer on the second side of the wafer by applying a dopant, exemplarily phosphorous, and diffusing the dopant into the wafer. Afterwards, a second dopant, exemplarily phosphorous or arsenic, may be applied either by directly applying the second dopant on the second side of the wafer or by applying the second dopant in the oxide layer of the carrier wafer 3 or of the wafer 1 . A third dopant, exemplarily boron, may be applied on the second side or in the oxide layer of the carrier wafer 3 or wafer 1 for forming the anode layer (in case of creating an IGBT). The second and third dopant may be diffused into the wafer in the same heat treatment step, which is performed for the creation for the front sided layers afterwards. All area implant shall mean that on a side of the wafer (front or backside) ions are implanted over the whole side. A partial implant shall mean that ions are implanted over a part of the side, as it is e.g. the case for n doped source regions which are laterally limited, Lateral is meant in view of the front or back side so that the source regions are present in a part of the front side, whereas there are other parts on the front side, in which no source regions are arranged.

The first wafer 2 may be a Float Zone FZ wafer. A FZ wafer is produced by a zone melting process also known as zone refining or floating zone process, in which a narrow region of a crystal is molten, and this molten zone is moved along the crystal, thus producing a highly purified crystal. The carrier wafer 3 may be a Czochralski wafer. A Czochralski wafer is produced by a Czochralski process, i.e. by pulling monocrystals against gravity out of a melt which is held in a crucible. Czochralski wafers are produced much more cost effective than FZ wafers, however, they have a higher impurity concentration than FZ wafers.

The semiconductor wafer 1 and the carrier wafer 3 form a stack, e.g. with a thickness of 700 Mm. The stack is then easily handled in further topology production processing. The stack is less likely to break than the thinned first wafer 2 (i.e. wafer 1 ) on its own. In addition, the stack has a higher mechanical yield. Further processing of the stack includes front end of line processing such as fabrication of individual devices, forming of n-type and p-type layers, passivation, metallization etc., to prepare the front side of the device until the final grinding step, i.e. the side opposite to the bonding side of the semiconductor wafer.

In a later/final phase of device production the carrier wafer 3 is removed, e.g. using mechanical grinding or chemical etching or a combination of those. In addition, a Back Grinding BG tape may be used to prevent wafer bowing after grinding. In one embodiment of the invention, the oxide layer at the bonding interface may serve as an electrical control end point. For instance while grinding the resistance of the surface may be monitored. When grinding process reaches the oxide layer a sudden change of resistance will occur. In addition, the oxide layer may be further etched off without etching the first wafer 2 or grinding may penetrate into the anode layer (in case of an IGBT) or the buffer region. This way the removal of the carrier wafer 3 may be controlled. Fig. 4 shows the semiconductor wafer 1 with a removed carrier wafer 3. All topology or device production processing such as front-end-of line processing on side 5 is carrier out before this point. At this stage of the device production back end of line processing may take place, e.g. back side metal stacking, sintering, annealing, etc., i.e. general standard device production for an insulated-gate bipolar transistor IGBT or a diode. For an IGBT, anode implanting and diffusion may be performed before creating a back side electrode.

Topology production processing on the first side comprises creation of a p doped layer (p base layer for an IGBT, p anode layer for a diode), which is exemplarily created by applying ions (i.e. implanting or deposing) on the front side, which are afterwards diffused into the wafer 1 . For power semiconductor devices as produced by the method disclosed above, the diffusion is performed at a temperature of at least 900 °C, exemplarily of at least 950 °C and exemplarily of at least 1250 °C. For a carrier wafer, which has been treated before by applying ions into the carrier wafer 3, such n doped second and/or p doped third ions are now diffused into the wafer 1 . For the manufacturing of an insulated gate bipolar transistor (IGBT) further layers are created on the first side like an n+ doped source region and a gate electrode in form of a planar or trench gate electrode. The creation of the main electrodes on the first and second side may be performed simultaneously, but the electrode on the first side may also be created before the destructive removal of the carrier wafer. The main electrode on the second side is created after the destructive removal such that the p doped anode layer (IGBT) or an n doped layer in from of a cathode layer (diode) contact the main electrode on the second side.

The term IGBT shall also cover a reverse conducting IGBT, in which on the second side the p doped IGBT anode layer alternates with n doped short regions.

In another embodiment, the conductivity types are switched, i.e. all layers of the n (first) conductivity type are p type (e.g. the drift layer) and all layers of the p (second) conductivity type are n type (e.g. IGBT anode layer). The first conductivity type is always different from the second conductivity type.

LIST OF DESIGNATIONS

1 Semiconductor wafer

2 First wafer

3 Carrier wafer

4 Oxide layer

5 Side

6 IGBT Anode

Claims

PATENT CLAIMS

A method for manufacturing a semiconductor device, in particular a power semiconductor device, comprising a semiconductor wafer (1 ) having a first and a second side opposite to the first side as a device element, wherein the method is comprised of at least the following steps: initial thinning of a first wafer (2) forming the semiconductor wafer (1 ) to a final thickness of the semiconductor wafer (1 );

performing a permanent bonding on the second side of the semiconductor wafer (1 ) to a carrier wafer (3);

topology production processing on the first side; and

finally destructive removing of the carrier wafer (3).

A method for manufacturing a semiconductor device according to claim 1 , wherein at least one of the first wafer (2) is a Float Zone wafer or the carrier wafer (3) is a Czochralski wafer.

A method for manufacturing a semiconductor device according to any of the claims 1 or

2, wherein the first wafer (2) is a non-prediffused wafer.

A method for manufacturing a semiconductor device according to any of the claims 1 to

3, wherein the method comprises further partial or all area applying ions on the semiconductor wafer (1 ) on the second side before bonding of the semiconductor wafer (1 ) to the carrier wafer (3) for a creation of a buffer layer.

A method for manufacturing a semiconductor device according to any of the previous claims, wherein at least one of the carrier wafer (3) or the semiconductor wafer (1 ) comprises an oxide layer at least on a side to be bonded, wherein the oxide layer servers further as a solid barrier for impurities of the carrier wafer (3) during topology production processing.

6. A method for manufacturing a semiconductor device according to any of the claims 1 to 4, wherein the method comprises further producing an implanted carrier wafer (3) by partial or all area implanting of the carrier wafer (3) on the side to be bonded.

7. A method for manufacturing a semiconductor device according to claim 6, wherein the 5 implanted carrier wafer (3) serves as a dopant source for the backside of the first wafer

(2) for the creation of at least one of a buffer layer or an anode layer.

8. A method for manufacturing a semiconductor device according to claim 7, wherein the method comprises further partial or all area applying ions on the semiconductor wafer (1 ) on the second side before bonding of the semiconductor wafer (1 ) to the carrier wafer o (3) for a creation of a first buffer layer and

wherein the implanted carrier wafer (3) serves as a dopant source for the backside of the first wafer (2) for the creation of a second buffer layer.

9. A method for manufacturing a semiconductor device according to any of the claims 1 to 8, wherein the destructive removing of the carrier wafer (3) is stopped at the bonding5 interface by an electrical control end point.

10. A method for manufacturing a semiconductor device according to any of the previous claims, wherein the method comprises further front-end of line FEOL and back-end of line BEOL processing.

1 1 . A method for manufacturing a semiconductor device according to any of the previous0 claims, wherein the semiconductor device is a bipolar transistor, in particular an insulated-gate bipolar transistor IGBT or a reverse-conducting IGBT or a diode.

12. A method for manufacturing a semiconductor device according to any of the previous claims, wherein performing the permanent bonding on the second side of the semiconductor wafer (1 ) to a carrier wafer (3) comprises a bond anneal step at a5 temperature between 300 °C to 800 °C.

13. A method for manufacturing a semiconductor device according to any of the previous claims, wherein the final thickness of the semiconductor wafer (1 ) is between 100 to 600 Mm.

14. A method for manufacturing a semiconductor device according to any of the previous claims, wherein the topology production processing on the first side comprising a heat treatment step with a temperature of at least 900 °C, in particular of at least 1250 °C.

15. A method for manufacturing a semiconductor device according to claim 8, wherein the topology production processing on the first side comprising a heat treatment step with a temperature of at least 900 °C, in particular of at least 1250 °C and wherein the ions from the implanted carrier wafer (3) serving as a dopant source are diffused into the wafer at the heat treatment step.