Semiconductor chip and semiconductor device

The present invention relates to a semiconductor device for performing signal transmission between chips or between a chip and a package and the semiconductor chips thereof and, more particularly, to a semiconductor device for contactlessly performing signal transmission among a plurality of chips wherein a plurality of chips are stacked in a layered manner (three-dimensional manner) and the semiconductor chips thereof.

In order to meet demands for increasing the speed and capacity of data transmission among a plurality of semiconductor chips and for reducing the packaging area of a semiconductor device, a variety of semiconductor devices wherein a plurality of semiconductor chips are stacked have been realized. Of these semiconductor devices, there is known a semiconductor device having a configuration in which electrodes formed on a chip are closely arranged contactlessly and oppositely to each other and inter-chip signal transmission is performed by means of capacitive coupling between the two electrodes. In this configuration, contact failure that often occurs when electrodes are brought into contact with each other, presents no problem. For this reason, this configuration is expected to come into practical use for the connection between semiconductor chips that requires multitude of electrodes, particularly among a plurality of CPUs, between a CPU and a memory, and among a plurality of memories, and for a semiconductor device in which these semiconductor chips are stacked.

FIG. 1 illustrates a typical semiconductor device for performing contactless signal transmission, wherein electrode A and electrode B are formed on the circuit surfaces of chip A and chip B, respectively, adjacently oppositely to each other. When electrode A is assumed to be a driver and electrode B is assumed to be a receiver, then electrostatic capacitance is present between the two electrodes, causing the two electrodes to capacitively couple with each other. That is, electrical lines of force generated from electrode A at this time reach electrode B, thereby placing the two electrodes in a state of being electrically coupled with each other (capacitive coupling). Consequently, a signal exchange is made from electrode A to electrode B.

In this related art, however, signal transmission among three or more chips is not possible, though signal transmission between opposed chips, i.e., one-to-one inter-chip transmission, is possible. When a signal is transmitted from lowermost-layer chip A to uppermost-layer chip C in cases where conventional chips are simply stacked in a three-dimensional manner as illustrated in FIG. 2 , assuming, for example, that the chips are configured using a commonly-used semiconductor substrate (for example, a silicon substrate), then electric fields transmitted from the lowermost layer to the uppermost layer attenuate due to the effect of conductive property (loss) that the substrate of chip B located in an intermediate layer has. Therefore, it has been difficult to adequately perform signal transmission between chip A and chip C.

Hence, in order to enable a signal transmission among three or more chips, Patent documents 1 and 2 disclose a stacked electrical circuit in which a conductive region for penetrating through the front and back sides of a silicon substrate is provided and a signal transmission is performed with the mode thereof categorized into (1) a case where a signal from a wafer stacked on an upper side is received; (2) a case where a signal is transmitted to a wafer stacked on an upper side; (3) a case where a signal from a wafer stacked on a lower side is received; and (4) a case where a signal is transmitted to a wafer stacked on a lower side.

In addition, Patent document 3 discloses a three-layer modular electronic system, though the system is intended not for signal transmission among three or more chips but for signal transmission between two chips. In the system, electrodes are provided on upper and lower sides between the chips and a dielectric material substrate, in which an interconnection that penetrates the electrodes is formed, is sandwiched between the electrodes.

Patent document 1: Japanese Patent Laid-Open No. 56-002662 Patent document 2: Japanese Patent Laid-Open No. 62-020362 Patent document 3: Published Japanese translations of PCT international publication No. 09-504908

Problems to be Solved by the Invention

Patent documents 1 and 2 do not disclose any specific exemplary embodiments with regard to circuits for performing transmission/reception and switching signals between the upper and lower sides. Accordingly, although a contactless method is employed, theses patent documents could not achieve an increase in the speed and capacity of inter-chip data transmission and a reduction in the packaging size of a semiconductor device. In addition, although Patent document 3 illustrates in FIG. 8 an embodiment in which a one-to-many connection is made between chips by using an interconnection within the dielectric material substrate, the exemplary embodiment suffers from a problem in which the operation speed decreases because a circuit of branch destination serves as a load on the circuit operation.

It is an object of the present invention to provide a semiconductor device and a semiconductor chip, which are constructed to have a high reliable interconnection by stacking three or more layers of chips without contacts therebetween, wherein for the semiconductor chip and the semiconductor device, there is a particularly strong demand to increase the speed and capacity of data transmission among a plurality of CPUs, between a CPU and a memory, and among a plurality of memories and the like, and there is a particularly strong demand to reduce the size of the semiconductor device.

Means to Solve the Problems

A semiconductor chip according to the present invention includes a silicon substrate on which a first signal transmission circuit and a first changeover switch are formed; a wiring layer formed on the silicon substrate; and a first capacitive-coupling upper electrode formed on the wiring layer; wherein a first capacitive-coupling lower electrode is additionally formed on the rear surface of the silicon substrate through a first via hole penetrating the silicon substrate, the first capacitive-coupling upper electrode is directly connected to the first signal transmission circuit, and the first capacitive-coupling lower electrode is connected to the first signal transmission circuit through the first via hole and the first changeover switch.

A semiconductor device according to the present invention comprises a semiconductor chip including a silicon substrate on which a first signal transmission circuit and a first changeover switch are formed; an interconnection layer formed on the silicon substrate; and a first capacitive-coupling upper electrode formed on the interconnection layer; wherein the first capacitive-coupling lower electrode is additionally formed on the rear surface of the silicon substrate through a first via hole that penetrates the silicon substrate, the first capacitive-coupling upper electrode is directly connected to the first signal transmission circuit, and the first capacitive-coupling lower electrode is connected to the first signal transmission circuit through the first via hole and the first changeover switch.

According to the present invention, in a semiconductor device comprising three or more layers of semiconductor chips, a signal transmission between respective two semiconductor chips is controlled and a signal transmission between predetermined electrodes within predetermined semiconductor chips is realized. In addition, since a signal is transmitted through a capacitive coupling, there is no need to ensure electrical conductivity as in the case of contact-type electrodes, thereby increasing the tolerance of positioning accuracy between semiconductor chips.

FIG. 1 is a cross-sectional view illustrating the configuration of a related semiconductor device for performing signal transmission by means of capacitive coupling between two chips.

FIG. 2 is a cross-sectional view illustrating the configuration of a related semiconductor device for performing signal transmission by means of capacitive coupling among three chips.

FIG. 3 is a schematic view illustrating a semiconductor device according to an exemplary embodiment of the present invention in which three semiconductor chips are stacked.

FIG. 4 is an enlarged view of an electrode pad portion in FIG. 3 , illustrating the configuration of a first exemplary embodiment.

FIG. 5 is a circuit diagram of semiconductor chip 2 b illustrated in FIG. 4 .

FIG. 6 is another enlarged view of the electrode pad portion in FIG. 3 , illustrating the configuration of a second exemplary embodiment.

FIG. 7 is a circuit diagram of semiconductor chip 2 b illustrated in FIG. 6 .

FIG. 8 is another enlarged view of the electrode pad portion in FIG. 3 , illustrating the configuration of a third exemplary embodiment.

FIG. 9 is a circuit diagram of semiconductor chip 2 b illustrated in FIG. 8 .

FIG. 10 is another enlarged view of the electrode pad portion in FIG. 3 , illustrating the configuration of a fourth exemplary embodiment.

FIG. 11 is a circuit diagram of semiconductor chip 2 b illustrated in FIG. 10 .

FIG. 12 is another enlarged view of the electrode pad portion in FIG. 3 , illustrating the configuration of a fifth exemplary embodiment.

FIG. 13 is a circuit diagram of semiconductor chip 2 b illustrated in FIG. 12 .

FIG. 14 is a circuit diagram of semiconductor chip 2 b of a sixth exemplary embodiment.

FIG. 15 is a schematic view to explain the operation of one-way communication in which the flow of data transmission is in a direction from an upper electrode to a lower electrode and a capacitive-coupling portion includes an amplifier element.

FIG. 16 is a schematic view to explain the operation of one-way communication in which the flow of data transmission is in a direction from an upper electrode to a lower electrode and a capacitive-coupling portion includes an amplifier fitted with a switch function.

FIG. 17 is a schematic view to explain the operation of two-way communication in which the flow of data transmission is in a direction from an upper electrode to a lower electrode and from the lower electrode to the upper electrode and a capacitive-coupling portion includes an amplifier element.

FIG. 18 is a schematic view to explain the operation of two-way communication in which the flow of data transmission is in a direction from an upper electrode to a lower electrode and from the lower electrode to the upper electrode and a capacitive-coupling portion includes an amplifier having a switch function.

FIG. 19 is a diagram of a transmitter/receiver circuit of the first exemplary embodiment.

FIG. 20 is a schematic view illustrating a timing chart of a transmitter/receiver circuit of the first exemplary embodiment.

FIG. 21 is a block diagram illustrating a sampling latch-type receiver circuit as an example of a receiver circuit of the first exemplary embodiment.

1 semiconductor device

2 a , 2 b , 2 c semiconductor chip

3 a , 3 b , 3 c silicon substrate

4 a , 4 b , 4 c interconnection layer

5 a , 5 b adhesion layer

6 a , 6 b , 6 c , 6 d , 6 e , 6 f transmitter/receiver circuit

7 , 7 a , 7 b , 7 c , 7 d , 11 a , 11 b , 11 c , 11 d , 12 a , 12 b , 12 c , 12 d , 12 e , 12 f , 12 g , 12 h , 7 A, 7 B, 7 C, 7 D, 7 A 1 , 7 B 1 , 7 C 1 , 7 A 2 , 7 B 2 , 7 C 2 , 17 A, 17 B, 17 C switch

8 a , 8 b , 8 c , 8 d , 8 e , 8 f , 8 g , 8 h electrode pad

9 , 9 a , 9 b through-via hole

10 terminal leading to internal circuit

10 A contact point

13 A, 13 B, 13 C, 13 D, 15 A 1 , 15 B 1 , 15 C 1 , 15 A 2 , 15 B 2 , 15 C 2 amplifier

14 A, 14 B, 14 C, 14 D, 16 A 1 , 16 B 1 , 16 C 1 , 16 A 2 , 16 B 2 , 16 C 2 amplifier with switch

18 transmission buffer

19 reception buffer, sampling latch

Exemplary embodiments of a semiconductor device according to the present invention will be described in detail with reference to the drawings.

FIG. 3 illustrates semiconductor device 1 in which three semiconductor chips 2 a , 2 b and 2 c are stacked. An electrode pad is formed on a surface of each semiconductor chip and these electrode pads are positioned opposite to each other at a certain distance, so that an inter-chip signal exchange is performed between the opposed electrode pads. In semiconductor chips 2 a and 2 c , an electrode pad is formed on a surface in which a circuit is formed, while in semiconductor chip 2 b , an electrode pad is also formed on the rear surface thereof. This semiconductor chip 2 b makes it possible to exchange a signal among semiconductor chips even if three or more semiconductor chips are stacked.

First Exemplary Embodiment of the Present Invention

FIG. 4 is an enlarged view of an electrode pad portion in FIG. 3 , and illustrates the configuration of a first exemplary embodiment. Semiconductor chip 2 a is configured such that at least one transmitter/receiver circuit 6 a is formed in silicon substrates 3 a , and electrode pad 8 a is formed in or on interconnection layers 4 a formed on the transmitter/receiver circuit. Semiconductor chip 2 c is configured such that at least one transmitter/receiver circuit 6 c is formed in silicon substrates 3 c , and electrode pad 8 d is formed in or on interconnection layers 4 c formed on the transmitter/receiver circuit. Semiconductor chip 2 b is configured such that in addition to at least one transmitter/receiver circuit 6 b , switch 7 is formed in silicon substrate 3 b . In addition, electrode pad 8 b is formed in or on interconnection layer 4 b and electrode pad 8 c is additionally formed on the rear surface of silicon substrate 3 b through via hole 9 that penetrates silicon substrate 3 b.

In semiconductor chip 2 b , electrode pad 8 b present on the circuit surface thereof is directly connected to transmitter/receiver circuit 6 b , whereas electrode pad 8 c present on the rear surface is connected to the transmitter/receiver circuit through via hole 9 and switch 7 .

Operation of the present invention will be described with reference to FIG. 5 .

FIG. 5 illustrates the circuit diagram of semiconductor chip 2 b previously shown. When a signal is input from electrode pad 8 b , switch 7 is opened. As a result, the signal that is input from electrode pad 8 b is supplied to transmitter/receiver circuit 6 b . On the other hand, when a signal from the transmitter/receiver circuit is input to electrode pad 8 c , switch 7 is closed. Although each of transmitter/receiver circuits 6 a , 6 b and 6 c is shown being integral as a transmitter/receiver circuit in the exemplary embodiment, the transmitter/receiver circuit may be a transmitter circuit and a receiver circuit that are separate from each other.

As can be seen from the circuit diagram of FIG. 5 , via hole 9 is connected to electrode pad 8 c . When a via hole is formed on a silicon substrate, a thin insulating layer must be provided on a boundary face between the silicon and the via hole, unlike when a via hole is formed on an organic substrate. Consequently, the parasitic capacitance of the via hole may amount to as large as several pF, which value is not negligible, as compared with capacitance formed between electrode pads of adjacent semiconductor chips. For this reason, in the present invention, switch 7 is provided in order to make the parasitic capacitance of this via hole 9 invisible from a signal that is input from electrode pad 8 b.

On the other hand, when a signal is output from transmitter/receiver circuit 6 b to electrode pad 8 c , electrode pad 8 b , as well as via hole 9 and electrode pad 8 c , are visible from transmitter/receiver circuit 6 b . However, the parasitic capacitance of electrode pad 8 b is sufficiently small compared with that of via hole 9 and hence is negligible.

Interposing switch 7 not only between transmitter/receiver circuit 6 b and electrode pad 8 c but also between transmitter/receiver circuit 6 b and electrode pad 8 b is conceivable. In the present exemplary embodiment, in order to achieve low power consumption, a switch is provided only in a portion that connects to the via hole and that has the largest parasitic capacitance.

When a signal is input from electrode pad 8 c , switch 7 is closed. As a result, the signal that is input from electrode pad 8 c is supplied to transmitter/receiver circuit 6 b . On the other hand, when a signal is output from the transmitter/receiver circuit to electrode pad 8 b , switch 7 is opened.

Semiconductor chips 2 a , 2 b and 2 c described above may be stacked as illustrated in FIG. 3 to form semiconductor device 1 .

Second Exemplary Embodiment of the Present Invention

FIG. 6 illustrates a second exemplary embodiment.

In the first exemplary embodiment, switch 7 is provided between transmitter/receiver circuit 6 b and via hole 9 . In the present exemplary embodiment, however, switch 7 b is additionally provided between transmitter/receiver circuit 6 b and electrode pad 8 b . Note that the switch provided between transmitter/receiver circuit 6 b and via hole 9 is denoted as switch 7 a . Both switches 7 a and 7 b are formed on the same silicon substrate as transmitter/receiver circuit 6 b . The remainder of the second exemplary embodiment is the same as the first exemplary embodiment described earlier.

Operation of the second exemplary embodiment will be described with reference to FIG. 7 . When a signal is input from electrode pad 8 b , switch 7 b is closed and switch 7 a is opened. As a result, via hole 9 having large parasitic capacitance is invisible from electrode pad 8 b . On the other hand, when a signal is output from transmitter/receiver circuit 6 b to electrode pad 8 c , switch 7 b is opened and switch 7 a is closed. As a result, the parasitic capacitance of electrode pad 8 b becomes invisible.

When a signal is input from electrode pad 8 c , switch 7 a is closed and switch 7 b is opened. As a result, the parasitic capacitance of electrode pad 8 b becomes invisible. On the other hand, when a signal is output from transmitter/receiver circuit 6 b to electrode pad 8 b , switch 7 a is opened and switch 7 b is closed. As a result, via hole 9 having large parasitic capacitance is invisible from electrode pad 8 b.

When a signal is input from electrode pad 8 c , the parasitic capacitance of electrode pad 8 b , as well as the parasitic capacitance of via hole 9 , is visible in the first exemplary embodiment. Accordingly, there is a possibility, particularly when chips are connected in a multistage manner beyond electrode pad 8 b , that this parasitic capacitance becomes larger, degrading signal quality. In the present exemplary embodiment, however, switch 7 b has the capability to eliminate the parasitic capacitance of electrode pad 8 b , thus improving the signal quality.

Semiconductor chips 2 a , 2 b and 2 c described above may be stacked as illustrated in FIG. 3 to form semiconductor device 1 .

Third Exemplary Embodiment of the Present Invention

FIG. 8 illustrates a third exemplary embodiment.

In the second exemplary embodiment, switch 7 a is provided between transmitter/receiver circuit 6 b and via hole 9 and switch 7 b is additionally provided between transmitter/receiver circuit 6 b and electrode pad 8 b . In the present exemplary embodiment, however, there is provided switch 7 c for directly opening/closing a path between upper and lower electrode pads 8 b and 8 c . All of switches 7 a , 7 b and 7 c are formed on the same silicon substrate as transmitter/receiver circuit 6 b . The remainder of the third exemplary embodiment is the same as the second exemplary embodiment described earlier.

FIG. 9 illustrates the circuit diagram shown in FIG. 8 .

A signal exchange between electrode pad 8 b and transmitter/receiver circuit 6 b and a signal change between electrode pad Sc and transmitter/receiver circuit 6 b are performed in the same manner as in the second exemplary embodiment. When a signal is directly transmitted between electrode pad 8 b and electrode pad 8 c , switch 7 c is closed and switches 7 a and 7 b are opened. In this case, transmitter/receiver circuit 6 b is not visible as a load.

The same signal transmission as described above is possible in the first and second exemplary embodiments. In that case, since the transmitter/receiver circuit is visible as a load, signal attenuation becomes accordingly large.

In the present exemplary embodiment, particularly when a signal needs to be directly transmitted from semiconductor chip 2 a to semiconductor chip 2 c or from semiconductor chip 2 c to semiconductor chip 2 a , the signal does not pass through two switches 7 a and 7 b , but is allowed to pass through only one switch 7 c . Therefore, signal attenuation becomes smaller.

Semiconductor chips 2 a , 2 b and 2 c described above may be stacked as illustrated in FIG. 3 to form semiconductor device 1 .

Fourth Exemplary Embodiment of the Present Invention

FIG. 10 illustrates a fourth exemplary embodiment.

Semiconductor chip 2 a is configured such that at least one transmitter/receiver circuit 6 a is formed in silicon substrate 3 a , and electrode pad 8 a is formed in or on interconnection layers 4 a formed on the transmitter/receiver circuit. Semiconductor chip 2 c is configured such that at least one transmitter/receiver circuit 6 c is formed in silicon substrate 3 c , and electrode pad 8 d is formed in or on interconnection layers 4 c formed on the transmitter/receiver circuit. Semiconductor chip 2 b is configured such that in addition to at least one transmitter/receiver circuit 6 b , switches 7 a and 7 d are formed in silicon substrate 3 b . Electrode pad 8 b is formed in or on interconnection layer 4 b , and electrode pad 8 c is additionally formed on the rear surface of silicon substrate 3 b through via hole 9 a that penetrates silicon substrate 3 b . Electrode pad 8 e is additionally formed on the rear surface of silicon substrate 3 b through via hole 9 b that penetrates silicon substrate 3 b.

In semiconductor chip 2 b , electrode pad 8 b that is present on the circuit surface thereof is directly connected to transmitter/receiver circuit 6 b , but electrode pad 8 c that is present on the rear surface is connected to transmitter/receiver circuit 6 b through via hole 9 a and switch 7 a , and electrode pad 8 e that is present on the rear surface is connected to transmitter/receiver circuit 6 b through via hole 9 b and switch 7 d.

Two switches 7 a and 7 d have the function of cutting off the interconnection so that the parasitic capacitances of via hole 9 a and 9 b are invisible from transmitter/receiver circuit 6 b and also have the function of selecting electrode pad 8 c or electrode pad 8 e.

FIG. 11 is the circuit diagram shown in FIG. 10 .

When switch 7 a is closed and switch 7 d is opened, a signal exchange between electrode pad 8 b and transmitter/receiver circuit 6 b and a signal change between electrode pad 8 c and transmitter/receiver circuit 6 b are performed in the same manner as in the first exemplary embodiment.

When switch 7 a is opened and switch 7 d is closed, a signal exchange between electrode pad 8 b and transmitter/receiver circuit 6 b and a signal exchange between electrode pad 8 e and transmitter/receiver circuit 6 b are performed in the same manner as in the first exemplary embodiment.

In the present exemplary embodiment, the data processing rate is increased by forming a one-to-many connection between the upper and lower chips. In addition, by providing switches 7 a and 7 d , loads on electrodes 8 c and 8 e can be isolated, thus maintaining high signal quality.

Semiconductor chips 2 a , 2 b and 2 c described above may be stacked as illustrated in FIG. 3 to form semiconductor device 1 . This semiconductor device is a semiconductor device wherein a one-to-many connection is formed between the upper and lower chips, and provides the above-described advantages.

Fifth Exemplary Embodiment of the Present Invention

FIG. 12 illustrates a fifth exemplary embodiment.

Semiconductor chip 2 a is configured such that at least one transmitter/receiver circuit 6 a is formed in silicon substrate 3 a , and electrode pad 8 a is formed in or on interconnection layer 4 a formed on the transmitter/receiver circuit. Semiconductor chip 2 c is configured such that at least one transmitter/receiver circuit 6 c is formed in silicon substrate 3 c , and electrode pad 8 d is formed in or on interconnection layer 4 c formed on the transmitter/receiver circuit. Semiconductor chip 2 b is configured such that in addition to at least one transmitter/receiver circuit 6 b , switches 11 a and 11 d are formed in silicon substrate 3 b . Furthermore, electrode pad 8 b is formed in or on interconnection layer 4 b , and electrode pad 8 c is additionally formed on the rear surface of silicon substrate 3 b through via hole 9 a that penetrates silicon substrate 3 b . Still furthermore, electrode pad 8 e is additionally formed on the rear surface of silicon substrate 3 b through via hole 9 b that penetrates silicon substrate 3 b.

In semiconductor chip 2 b , electrode pad 8 b that is present on the circuit surface thereof is directly connected to transmitter/receiver circuit 6 b , but electrode pad 8 c that is present on the rear surface is connected to transmitter/receiver circuit 6 b through via hole 9 a and switch 11 a . In addition, electrode pad 8 e that is present on the rear surface is connected to transmitter/receiver circuit 6 b through via hole 9 b and switch 11 d.

Two switches 11 a and 11 d have the function of cutting off the interconnection so that the parasitic capacitances of via hole 9 a and 9 b are invisible from transmitter/receiver circuit 6 b and also have the function of selecting electrode pad 8 c or electrode pad 8 e.

In addition, semiconductor chip 2 a is configured such that at least one transmitter/receiver circuit 6 e is formed in silicon substrate 3 a , and electrode pad 8 f is formed in or on interconnection layer 4 a formed on the transmitter/receiver circuit. Semiconductor chip 2 c is configured such that at least one transmitter/receiver circuit 6 f is formed in silicon substrate 3 c , and electrode pad 8 h is formed in or on interconnection layer 4 c formed on the transmitter/receiver circuit. Semiconductor chip 2 b is configured such that in addition to at least one transmitter/receiver circuit 6 d , switches 11 b and 11 c are formed in silicon substrate 3 b . Furthermore, electrode pad 8 g is formed in or on interconnection layer 4 b.

In semiconductor chip 2 b , electrode pad 8 g that is present on the circuit surface thereof is directly connected to transmitter/receiver circuit 6 d , but electrode pad 8 e that is present on the rear surface is connected to transmitter/receiver circuit 6 d through via hole 9 b and switch 11 b . In addition, electrode pad 8 c that is present on the rear surface is connected to transmitter/receiver circuit 6 d through via hole 9 a and switch 11 c.

Two switches 11 b and 11 c have the function of cutting off the interconnection so that the parasitic capacitances of via hole 9 a and 9 b are invisible from transmitter/receiver circuit 6 d and also have the function of selecting electrode pad 8 c or electrode pad 8 e.

That is, formed on semiconductor chip 2 b are two transmitter/receiver circuits 6 b and 6 d and four switches 11 a , 11 b , 11 c and 11 d . Thus, it is possible to select one electrode from upper, lower, left and right electrode pads 8 b , 8 c , 8 g and 8 e.

FIG. 13 illustrates the circuit diagram shown in FIG. 12 .

When switch 11 a is closed and switch 11 d is opened, a signal exchange between electrode pad 8 b and transmitter/receiver circuit 6 b and a signal change between electrode pad 8 c and transmitter/receiver circuit 6 b are performed in the same manner as in the first exemplary embodiment.

When switch 11 a is opened and switch 11 d is closed, a signal exchange between electrode pad 8 b and transmitter/receiver circuit 6 b and a signal exchange between electrode pad 8 e and transmitter/receiver circuit 6 b are performed in the same manner as in the first exemplary embodiment.

When switch 11 b is closed and switch 11 c is opened, a signal exchange between electrode pad 8 g and transmitter/receiver circuit 6 d and a signal exchange between electrode pad 8 e and transmitter/receiver circuit 6 d are performed in the same manner as in the first exemplary embodiment.

When switch 11 b is opened and switch 11 c is closed, a signal exchange between electrode pad 8 g and transmitter/receiver circuit 6 d and a signal exchange between electrode pad 8 c and transmitter/receiver circuit 6 d are performed in the same manner as in the first exemplary embodiment.

In the present exemplary embodiment, the data processing rate is increased higher than in the fourth exemplary embodiment by forming a many-to-many connection between the upper and lower chips. In addition, loads on electrodes 8 c and 8 e can be isolated by the switches, thus maintaining high signal quality.

Semiconductor chips 2 a , 2 b and 2 c described above may be stacked as illustrated in FIG. 3 to form semiconductor device 1 . This semiconductor device is a semiconductor device wherein a one-to-many connection is formed between the upper and lower chips form, and provides the above-described advantages.

Sixth Exemplary Embodiment of the Present Embodiment

FIG. 14 illustrates a circuit diagram of a sixth exemplary embodiment. The sixth exemplary embodiment is a combination of the third exemplary embodiment and the fifth exemplary embodiment.

Switch 12 b is provided between transmitter/receiver circuit 6 b and electrode pad 8 b and switch 12 g is provided between transmitter/receiver circuit 6 b and electrode pad 8 g . In addition, switch 12 h is provided between transmitter/receiver circuit 6 d and electrode pad 8 b and switch 12 d is provided between transmitter/receiver circuit 6 d and electrode pad 8 g.

Semiconductor chip 2 b is configured such that in addition to at least one transmitter/receiver circuit 6 b , switches 12 a and 12 f are formed in silicon substrate 3 b . Furthermore, electrode pad 8 b is formed in or on interconnection layer 4 b , and electrode pad 8 c is additionally formed on the rear surface of silicon substrate 3 b through via hole 9 a that penetrates silicon substrate 3 b . Still furthermore, electrode pad 8 e is additionally formed on the rear surface of silicon substrate 3 b through via hole 9 b that penetrates silicon substrate 3 b.

Electrode pad 8 c is connected to transmitter/receiver circuit 6 b through via hole 9 a and switch 12 a . In addition, electrode pad 8 e is connected to transmitter/receiver circuit 6 b through via hole 9 b and switch 12 f.

Two switches 12 a and 12 f have the function of cutting off the interconnection so that the parasitic capacitances of via hole 9 a and 9 b are invisible from transmitter/receiver circuit 6 b and also have the function of selecting electrode pad 8 c or electrode pad 8 e.

In addition, semiconductor chip 2 b is configured such that in addition to at least one transmitter/receiver circuit 6 d , switches 12 c and 12 e are formed on silicon substrate 3 b . Furthermore, electrode pad 8 g is formed in or on interconnection layer 4 b.

Electrode pad 8 e is connected to transmitter/receiver circuit 6 d through via hole 9 b and switch 12 c . In addition, electrode pad 8 c is connected to transmitter/receiver circuit 6 d through via hole 9 a and switch 12 e.

Two switches 12 c and 12 e have the function of cutting off the interconnection so that the parasitic capacitances of via hole 9 a and 9 b are invisible from transmitter/receiver circuit 6 d and also have the function of selecting electrode pad 8 c or electrode pad 8 e.

In the present exemplary embodiment, there are provided switches 12 h and 12 e for directly opening/closing a path between upper and lower electrode pads 8 b and 8 c and switches 12 g and 12 f for directly opening/closing a path between upper and lower electrode pads 8 g and 8 e.

That is, formed on semiconductor chip 2 b are two transmitter/receiver circuits 6 b and 6 d and eight switches 12 a , 12 b , 12 c , 12 d , 12 e , 12 f , 12 g and 12 h that form a crossbar switch. Thus, it is possible to select one electrode from upper, lower, left and right electrode pads 8 b , 8 c , 8 d and 8 e.

When switches 12 a , 12 b , 12 e and 12 h are closed and switches 12 c , 12 d , 12 f and 12 g are opened, a signal exchange between electrode pad 8 b and transmitter/receiver circuit 6 b and a signal exchange between electrode pad 8 c and transmitter/receiver circuit 6 b are performed in the same manner as in the third exemplary embodiment.

When switches 12 a , 12 b , 12 e and 12 h are opened and switches 12 c , 12 d , 12 f and 12 g are closed, a signal exchange between electrode pad 8 g and transmitter/receiver circuit 6 d and a signal exchange between electrode pad 8 e and transmitter/receiver circuit 6 d are performed in the same manner as in the third exemplary embodiment.

When switches 12 a , 12 c , 12 e and 12 f are closed and switches 12 b , 12 d , 12 g and 12 h are opened, a signal exchange between electrode pad 8 b and transmitter/receiver circuit 6 b , a signal exchange between electrode pad 8 c and transmitter/receiver circuit 6 b , a signal exchange between electrode pad 8 g and transmitter/receiver circuit 6 d and a signal exchange between electrode pad 8 e and transmitter/receiver circuit 6 d are performed in the same manner as in the fifth exemplary embodiment.

In the present exemplary embodiment, the data processing rate is increased higher than in the fifth exemplary embodiment by connecting between the upper and lower chips by means of crossbar switches. In addition, loads on a plurality of electrodes can be isolated by the switches, thus maintaining high signal quality.

Since the crossbar switch configuration allows a required contact point to be selected and then opened or closed, problems such as inter-chip data collision or crosstalk are less likely to occur and transmission speed can be easily increased. These effects are significant particularly when a plurality of microcontrollers are connected to one another in the CPU-to-CPU connection.

Semiconductor chips 2 a , 2 b and 2 c described above may be stacked as illustrated in FIG. 1 to form semiconductor device 1 . This semiconductor device is a semiconductor device wherein the upper and lower chips are connected by crossbar switches, and provides the above-described advantages.

Seventh Exemplary Embodiment of the Present Invention

When a plurality of chips are stacked and inter-chip data is transmitted and received through capacitive coupling, the following problems occur: 1) a signal to be transmitted/received has a differential waveform; 2) when a signal passes through a through-via hole, the signal is affected by the parasitic capacitance at the through-via hole and the strength of a received signal decreases; and 3) since a signal is transmitted through switches 12 a to 12 h shown in, for example, the sixth exemplary embodiment, the strength of the received signal is decreased by the resistances of switches 12 a to 12 h . A description will now be made of a method to solve these problems by using amplifier elements in data transmission/reception through capacitive coupling, with reference to FIG. 15 . FIG. 15 illustrates one-way communication in which data transmission is performed in a direction from an upper electrode to a lower electrode.

First, when data is transmitted from upper electrode 8 b to terminal 10 for transmitter/receiver circuit 6 b or the like, switches 7 A and 7 B are turned on and switches 7 C and 7 D are turned off. As a result, data that is input from upper electrode 8 b is received and amplified by amplifier 13 A, passes through switch 7 A and contact point 10 A and is received and amplified by amplifier 13 B. Data that is amplified by amplifier 13 B is input to an internal circuit through switch 7 B. At this time, amplifiers 13 C and 13 D receive and amplify similar data. However, since switches 7 C and 7 D are turned off, contention does not take place between data from amplifier 13 A and data from amplifier 13 C at contact point 10 A, and data transmission to lower electrode 8 c by way of through-via hole 9 also does not take place. As a result, a desired data transmission is performed without causing any false operation. Furthermore, by performing data transmission through amplifiers, it is possible to accurately receive a signal that is attenuated by capacitive coupling and the resistance of switch 7 A. In addition, when there are no amplifiers, transmission power needs to be increased in order to maintain the strength of a received signal to be greater than the sensitivity of a receiver circuit, but by performing data transmission through amplifiers, transmission power can be reduced.

Next, when data from the internal circuit is transferred to lower electrode 8 c , switches 7 C and 7 D are turned on and switches 7 A and 7 B are turned off. As a result, the data from the internal circuit is received and amplified by amplifier 13 C, passes through switch 7 C and contact point 10 A, and is then transmitted to amplifier 13 D. The transmitted data is received and amplified by amplifier 13 D, and output to lower electrode 8 c through switch 7 D and through-via hole 9 . At this time, since amplifier 13 B is connected to contact point 10 A, amplifier 13 B receives and amplifies similar data. However, since switch 7 B is turned off, there is no occurrence of data contention at terminal 10 that is connected to the internal circuit and no false operation is caused. In addition, although data that is input to upper electrode 8 b is received and amplified by amplifier 13 A, there is no occurrence of data contention at contact point 10 A and no operation is caused because switch 7 A is turned off. As a result, a desired data transmission is performed in a direction from the internal circuit toward lower electrode 8 c . Furthermore, since the signal is amplified by amplifier 13 D, it is possible to convey the signal received at upper electrode 8 b to lower electrode 8 c with the signal being amplified. Consequently, it is possible to transmit a signal to a lower chip with the signal being fully amplified, although the signal transmission is executed by means of capacitive coupling. When three or more chips are stacked, if there are no amplifiers, the transmission amplitude needs to be increased so that three or more chips can perform transmission. When data transmission is performed through amplifiers, however, the transmission amplitude may only be as large as is required for one chip, since a signal is amplified for transmission from each one chip to another. In addition, when there are no amplifiers, only the chips, the number of which is the number of chips that is determined at the time of design, can be stacked. When data transmission is performed through amplifiers, however, the number of chips to be stacked needs not be considered at the time of design.

When data from an upper electrode is transmitted to both the internal circuit and lower electrode 8 c , switches 7 A, 7 B and 7 D are turned on and switch 7 C is turned off. As a result, the data from upper electrode 8 b is transmitted to the internal circuit through amplifier 13 A, switch 7 A, amplifier 13 B and switch 7 B and, at the same time, is also transmitted to lower electrode 8 c through amplifier 13 A, switch 7 A, amplifier 13 D and switch 7 D. At this time, amplifier 13 C receives and amplifies similar data. However, since switch 7 C is turned off, no conflict occurs at contact point 10 A between the data from the upper electrode and the data from the internal circuit. Thus, the chips can operate normally.

An explanation will now be made of the operation of one-way communication in which data transmission is performed in a direction from an upper electrode to a lower electrode and a capacitive-coupling portion includes amplifiers with a switch function, with reference to FIG. 16 .

When a capacitive-coupling portion includes amplifier elements, as described in FIG. 15 , data transmission from upper electrode 8 b to lower electrode 8 c can also be achieved even when using amplifiers that have a switch function.

When data is transmitted from upper electrode 8 b to the internal circuit, for example, it is possible to perform the data transmission normally, while avoiding data contention, by turning on amplifiers 14 A and 14 B with a switch and turning off amplifiers 14 C and 14 D with a switch in the same way as in FIG. 15 . Data transmission from the internal circuit to lower electrode 8 c and data transmission from upper electrode 8 b to both the internal circuit and lower electrode 8 c can be performed normally, while avoiding data contention in the same manner as in FIG. 15 .

By connecting an amplifier in series with a switch or by using an amplifier with a switch, a reduction in amplitude due to the resistance of a switch can be improved thereby enhancing signal quality.

Eighth Exemplary Embodiment of the Present Invention

An explanation will now be made of operation in a case in which a flow of data is both in a direction from an upper electrode to a lower electrode and in a direction from a lower electrode to an upper electrode, and a capacitive-coupling portion includes amplifiers that have a switch function, with reference to FIG. 17 .

When data is transmitted from upper electrode 8 b to an internal circuit, switches 7 A 1 and 7 B 1 are turned on, and switches 7 A 2 , 7 B 2 , 7 C 1 and 7 C 2 are turned off. As a result, data that is input to upper electrode 8 b is received and amplified by amplifier 15 A 1 , passes through switch 7 A 1 and is received and amplified by amplifier 15 B 1 . The data that is amplified by amplifier 15 B 1 passes through switch 7 A 1 , and is transmitted to the internal circuit through switch 7 B 1 . At this time, since switches 7 A 2 , 7 B 2 , 7 C 1 and 7 C 2 are turned off, there is no data contention at contact point 10 A. Thus, the data can be transmitted normally. On the other hand, when data is transmitted from the internal circuit to upper electrode 8 b , data transmission can be performed, without causing any data contention, by turning on switches 7 B 2 and 7 A 2 and turning off switches 7 A 1 , 7 B 1 , 7 C 1 and 7 C 2 . Similarly, when data is transmitted from upper electrode 8 b to lower electrode 8 c or from lower electrode 8 c to upper electrode 8 b , data transmission can be performed, as in the case of transmitting data from upper electrode 8 b to lower electrode 8 c , by correctly selecting switches to be turned on and to be turned off. In addition, it is also possible to simultaneously transmit data from one electrode toward two electrodes, as in the case of transmission from upper electrode 8 b to both the internal circuit and lower electrode 8 c.

Ninth Exemplary Embodiment of the Present Invention

FIG. 18 illustrates a ninth exemplary embodiment. In the present exemplary embodiment, amplifier elements that have a switch are used in place of the amplifiers and switch elements that are separately arranged in the eighth exemplary embodiment. Since amplifier elements fitted with a switch are used in place of amplifiers and switches, the present exemplary embodiment is the same in operation as the eighth exemplary embodiment, but has advantages such as areal reductions and reductions in power consumption.

Examples of Transmitter/Receiver Circuit of First Exemplary Embodiment

Examples of a transmitter/receiver circuit of the first exemplary embodiment will be described using the transmitter/receiver circuit diagram of FIG. 19 and the timing chart of FIG. 20 .

Transmission data is a non-return-to-zero (NRZ) signal which is transmitted in synchronization with the rising timing of a clock signal and is input from terminal Din to transmission buffer 18 . The level of a signal to be conveyed to lower electrode 8 c by way of through-via hole 9 is determined to be either the voltage of data or half the voltage of data, according to transmission clock CLK. That is, the transmission data of a transmission buffer output is conveyed to lower electrode 8 c within a high-level period of transmission clock CLK, and a voltage level half the voltage of the transmission data is conveyed to lower electrode 8 c within a low-level period of transmission clock CLK. Only the information on transition of the data at lower electrode 8 c is conveyed to upper electrode 8 d coupled with lower electrode 8 c through capacitive coupling, due to the capacitive coupling. For example, when the transmission data is “1”, lower electrode 8 c is set to “1” at the rising timing of transmission clock CLK and an upward-going pulse is conveyed to upper electrode 8 d according to the data transition of lower electrode 8 c from “½” to “1”. Lower electrode 8 c is set to “½” at the falling timing of transmission clock CLK and a downward-going pulse is conveyed to upper electrode 8 d according to the data transition of lower electrode 8 c from “1” to “½”. Conversely, when the transmission data is “0”, lower electrode 8 c is set to “0” at the rising timing of transmission clock CLK, and a downward-going pulse is conveyed to upper electrode 8 d according to the data transition of lower electrode 8 c from “½” to “0”. Lower electrode 8 c is set to “½” at the falling timing of transmission clock CLK and an upward-going pulse is conveyed to upper electrode 8 d according to the data transition of lower electrode 8 c from “0” to “½”. In this way, a transmission data of “1” or “0” is converted to an upward-going pulse or a downward-going pulse at upper electrode 8 d at the rising timing of transmission clock CLK. The transmission data is thus conveyed to upper electrode 8 d.

On the other hand, in a receiving-side circuit including upper electrode 8 d , switch 17 C operates according to reception clock CLK. When reception clock CLK is at a high level, the data of upper electrode 8 d is received by a reception buffer or by a sampling latch. When reception clock CLK is at a low level, a half voltage is applied to upper electrode 8 d , thereby resetting the voltage at the upper electrode.

FIG. 21 illustrates a sampling latch-type receiver circuit as an example of receiver circuit 19 .

A capacitive-coupling voltage is applied to terminal “in”, and a half voltage is applied to the other input terminal forming a differential pair together with terminal “in”. This sampling latch circuit operates in such a manner that when a CLK input is at a low level, all four wire connections of the drain terminals of nMOSs that are connected to wire connections “out” and “outb”, terminal “in” and the half-voltage input terminal are pre-charged to a power supply voltage. At this time, if the CLK input changes to a high level, then the receiver circuit senses a difference between voltages that are input to the “in” terminal and the half-voltage terminal and outputs the result of sensing as differential outputs “out” and “outb”.

Although the present invention has been described with reference to the exemplary embodiments, the present invention is not limited to the above-described exemplary embodiments. Various modifications that can be understood to those skilled in the art may be made to the constitution and details of the present invention within the scope thereof.

This application is the National Phase of PCT/JP2008/053612, filed Feb. 29, 2008, which claims the right of priority based on Japanese Patent Application No. 2007-060352, filed on Mar. 9, 2007, the entire content of which is incorporated herein by reference.