Semiconductor-element-including memory device

Abstract

A memory device includes pages each constituted by memory cells on a substrate. Voltages applied to first and second gate conductor layers and impurity layers in each memory cell are controlled to retain positive holes inside a channel semiconductor layer. In a page write operation, the voltage of the channel semiconductor layer is set to a first data retention voltage. In a page erase operation, the applied voltages are controlled to discharge the positive holes, and the voltage of the channel semiconductor layer is set to a second data retention voltage. At a second time after a first time, a memory re-erase operation is performed for the channel semiconductor layers at the second data retention voltage at the first time. At a third time after the second time, a memory re-write operation is performed for the channel semiconductor layers at the first data retention voltage at the first time.

Claims

1. A semiconductor-element-including memory device, the memory device comprising a plurality of pages arranged in a column direction, each of the pages being constituted by a plurality of memory cells arranged in a row direction on a substrate, each of the memory cells included in each of the pages comprising: a semiconductor base material that stands on the substrate in a vertical direction or that extends along the substrate in a horizontal direction; a first impurity layer and a second impurity layer that are disposed at respective ends of the semiconductor base material; a first gate insulating layer that partially or entirely surrounds a side surface of the semiconductor base material between the first impurity layer and the second impurity layer and that is in contact with or in close vicinity to the first impurity layer; a second gate insulating layer that surrounds the side surface of the semiconductor base material, that is connected to the first gate insulating layer, and that is in contact with or in close vicinity to the second impurity layer; a first gate conductor layer that partially or entirely covers the first gate insulating layer; a second gate conductor layer that partially or entirely covers the second gate insulating layer; and a channel semiconductor layer that is the semiconductor base material and that is covered by the first gate insulating layer and the second gate insulating layer, wherein voltages applied to the first gate conductor layer, the second gate conductor layer, the first impurity layer, and the second impurity layer are controlled to retain a group of positive holes, inside the channel semiconductor layer, generated by an impact ionization phenomenon, in a page write operation, a voltage of the channel semiconductor layer is made equal to a first data retention voltage that is higher than the voltage of either the first impurity layer or the second impurity layer or that is higher than the voltages of both the first impurity layer and the second impurity layer, in a page erase operation, the voltages applied to the first impurity layer, the second impurity layer, the first gate conductor layer, and the second gate conductor layer are controlled to discharge the group of positive holes through either the first impurity layer or the second impurity layer or both the first impurity layer and the second impurity layer, and the voltage of the channel semiconductor layer is made equal to a second data retention voltage that is lower than the first data retention voltage, the voltage of the channel semiconductor layer that is the semiconductor base material in each of the plurality of memory cells in each of the pages is equal to the first data retention voltage or the second data retention voltage at a first time, at a second time after a lapse of time since the first time, a memory re-erase operation is performed for the semiconductor base material of the channel semiconductor layer, in the pages, that is at a voltage equal to the second data retention voltage at the first time to perform a first refresh operation of returning the voltage of the channel semiconductor layer to a voltage close to the second data retention voltage, and at a third time after a lapse of time since the second time, a memory re-write operation is performed for the semiconductor base material of the channel semiconductor layer, in the pages, that is at a voltage equal to the first data retention voltage at the first time to perform a second refresh operation of returning the voltage of the channel semiconductor layer to a voltage close to the first data retention voltage.

2. The semiconductor-element-including memory device according to claim 1, wherein a first gate capacitance between the first gate conductor layer and the channel semiconductor layer is larger than a second gate capacitance between the second gate conductor layer and the channel semiconductor layer.

3. The semiconductor-element-including memory device according to claim 1, wherein the first gate conductor layer is isolated into at least two conductor layers around the first gate insulating layer when viewed in an axial direction of the semiconductor base material.

4. The semiconductor-element-including memory device according to claim 1, wherein the impact ionization phenomenon occurs inside the channel semiconductor layer between the first gate conductor layer and the second gate conductor layer to generate the group of positive holes inside the channel semiconductor layer.

5. The semiconductor-element-including memory device according to claim 1, wherein the first impurity layer is connected to a source line, the second impurity layer is connected to a bit line, one of the first gate conductor layer or the second gate conductor layer is connected to a word line, the other of the first gate conductor layer or the second gate conductor layer is connected to a first driving control line, and the word line and the first driving control line are connected to a row decoder circuit, the source line is connected to the semiconductor base materials in each block in the group of blocks, and voltages applied to the source line, the bit line, the first driving control line, and the word line are controlled to perform either the memory re-write operation or the memory re-erase operation or simultaneously perform both the memory re-write operation and the memory re-erase operation for all the semiconductor base materials in a block selected from among the group of blocks.

6. The semiconductor-element-including memory device according to claim 5, wherein in the first refresh operation and in the second refresh operation, an all-word-line selection signal and an all-plate-line selection signal are input to the row decoder circuit to select all the word lines and all the first driving control lines in the block.

7. The semiconductor-element-including memory device according to claim 5, wherein the first driving control line is arranged as a common driving control line for the memory cells adjacent to each other.

8. The semiconductor-element-including memory device according to claim 5, wherein the source line is isolated into source lines each of which is for the memory cells that are arranged in the column direction and which are disposed parallel to the word lines and the first driving control line.

9. The semiconductor-element-including memory device according to claim 5, wherein in a page sum-of-products read operation in which at least two word lines are selected in a multiple selection, the first refresh operation and the second refresh operation are performed in advance at least once for the word lines selected in a multiple selection in the page sum-of-products read operation.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 is a structural diagram of an SGT-including memory device according to a first embodiment;

(2) FIGS. 2A to 2C are diagrams for explaining an effect attained in a case where the gate capacitance of a first gate conductor layer connected to a plate line is made larger than the gate capacitance of a second gate conductor layer to which a word line is connected in the SGT-including memory device according to the first embodiment;

(3) FIGS. 3AA to 3AC are diagrams for explaining a mechanism of a write operation of the SGT-including memory device according to the first embodiment;

(4) FIG. 3B includes diagrams for explaining the mechanism of the write operation of the SGT-including memory device according to the first embodiment;

(5) FIG. 4A is a diagram for explaining a mechanism of a page erase operation of the SGT-including memory device according to the first embodiment;

(6) FIGS. 4BA to 4BD are diagrams for explaining the mechanism of the page erase operation of the SGT-including memory device according to the first embodiment;

(7) FIG. 4C includes diagrams for explaining the mechanism of the page erase operation of the SGT-including memory device according to the first embodiment;

(8) FIGS. 4DA to 4DD are diagrams for explaining a mechanism of the page erase operation of the SGT-including memory device according to the first embodiment;

(9) FIGS. 4EA to 4ED are diagrams for explaining a mechanism of the page erase operation of the SGT-including memory device according to the first embodiment;

(10) FIGS. 5A to 5C are diagrams for explaining a mechanism of a read operation of the SGT-including memory device according to the first embodiment;

(11) FIGS. 6AA and 6AB are diagrams for explaining a first refresh operation of the SGT-including memory device according to the first embodiment;

(12) FIG. 6B is an operation waveform diagram for explaining the first refresh operation of the SGT-including memory device according to the first embodiment;

(13) FIGS. 6CA and 6CB are diagrams for explaining a second refresh operation of the SGT-including memory device according to the first embodiment;

(14) FIG. 6D is an operation waveform diagram for explaining the second refresh operation of the SGT-including memory device according to the first embodiment;

(15) FIG. 6E is a circuit block diagram for explaining the first refresh operation and the second refresh operation of the SGT-including memory device according to the first embodiment;

(16) FIG. 6F is a circuit block diagram for explaining the first refresh operation and the second refresh operation of the SGT-including memory device according to the first embodiment;

(17) FIG. 6G is a circuit block diagram for explaining the first refresh operation of the SGT-including memory device according to the first embodiment;

(18) FIG. 6H is a circuit block diagram for explaining the second refresh operation of the SGT-including memory device according to the first embodiment;

(19) FIG. 6I is a circuit block diagram for explaining the first refresh operation and the second refresh operation of the SGT-including memory device according to the first embodiment;

(20) FIG. 6J is a circuit block diagram for explaining the first refresh operation and the second refresh operation of the SGT-including memory device according to the first embodiment;

(21) FIGS. 7A to 7D are diagrams for explaining a write operation of a DRAM memory cell including no capacitor in the related art;

(22) FIGS. 8A and 8B are diagrams for explaining a problem in the operation of the DRAM memory cell including no capacitor in the related art; and

(23) FIGS. 9A to 9C are diagrams for explaining a read operation of the DRAM memory cell including no capacitor in the related art.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

(24) Hereinafter, a semiconductor-element-including memory device (hereinafter called a dynamic flash memory) according to embodiments of the present invention will be described with reference to the drawings.

First Embodiment

(25) The structure and operation mechanisms of a dynamic flash memory cell according to a first embodiment of the present invention will be described with reference to FIG. 1 to FIGS. 5A to 5C. The structure of the dynamic flash memory cell will be described with reference to FIG. 1. An effect attained in a case where the gate capacitance of a first gate conductor layer 5a connected to a plate line PL is made larger than the gate capacitance of a second gate conductor layer 5b to which a word line WL is connected will be described with reference to FIGS. 2A to 2C. A mechanism of a data write operation will be described with reference to FIGS. 3AA to 3AC and FIG. 3B, mechanisms of a data erase operation will be described with reference to FIG. 4A to FIGS. 4EA to 4ED, and a mechanism of a data read operation will be described with reference to FIGS. 5A to 5C.

(26) FIG. 1 illustrates the structure of the dynamic flash memory cell according to the first embodiment of the present invention. On the top and the bottom of a silicon semiconductor pillar 2 (the silicon semiconductor pillar is hereinafter referred to as Si pillar) (which is an example of semiconductor base material in the claims) of the P or i (intrinsic) conductivity type formed on a substrate, N.sup.+ layers 3a and 3b (which are examples of first impurity layer and second impurity layer in the claims), one of which functions as the source and the other functions as the drain, are formed respectively. The part of the Si pillar 2 between the N.sup.+ layers 3a and 3b that function as the source and the drain functions as a channel region 7 (which is an example of channel semiconductor layer in the claims). Around the channel region 7, a first gate insulating layer 4a (which is an example of first gate insulating layer in the claims) and a second gate insulating layer 4b (which is an example of second gate insulating layer in the claims) are formed. The first gate insulating layer 4a and the second gate insulating layer 4b are in contact with or in close vicinity to the N.sup.+ layers 3a and 3b that function as the source and the drain respectively. Around the first gate insulating layer 4a and the second gate insulating layer 4b, the first gate conductor layer 5a (which is an example of first gate conductor layer in the claims) and the second gate conductor layer 5b (which is an example of second gate conductor layer in the claims) are formed respectively. The first gate conductor layer 5a and the second gate conductor layer 5b are isolated from each other by an insulating layer 6 (which is also referred to as first insulating layer). The channel region 7 between the N.sup.+ layers 3a and 3b is constituted by a first channel Si layer 7a surrounded by the first gate insulating layer 4a and a second channel Si layer 7b surrounded by the second gate insulating layer 4b. Accordingly, the N.sup.+ layers 3a and 3b that function as the source and the drain, the channel region 7, the first gate insulating layer 4a, the second gate insulating layer 4b, the first gate conductor layer 5a, and the second gate conductor layer 5b constitute a dynamic flash memory cell 10. The N.sup.+ layer 3a that functions as the source is connected to a source line SL (which is an example of source line in the claims), the N.sup.+ layer 3b that functions as the drain is connected to a bit line BL (which is an example of bit line in the claims), the first gate conductor layer 5a is connected to the plate line PL (which is an example of first driving control line in the claims), and the second gate conductor layer 5b is connected to the word line WL (which is an example of word line in the claims). Desirably, the structure is such that the gate capacitance of the first gate conductor layer 5a to which the plate line PL is connected is larger than the gate capacitance of the second gate conductor layer 5b to which the word line WL is connected.

(27) In FIG. 1, to make the gate capacitance of the first gate conductor layer 5a connected to the plate line PL larger than the gate capacitance of the second gate conductor layer 5b to which the word line WL is connected, the gate length of the first gate conductor layer 5a is made longer than the gate length of the second gate conductor layer 5b. Alternatively, instead of making the gate length of the first gate conductor layer 5a longer than the gate length of the second gate conductor layer 5b, the thicknesses of the respective gate insulating layers may be made different such that the thickness of the gate insulating film of the first gate insulating layer 4a is thinner than the thickness of the gate insulating film of the second gate insulating layer 4b. Alternatively, the dielectric constants of the materials of the respective gate insulating layers may be made different such that the dielectric constant of the gate insulating film of the first gate insulating layer 4a is higher than the dielectric constant of the gate insulating film of the second gate insulating layer 4b. The gate capacitance of the first gate conductor layer 5a connected to the plate line PL may be made larger than the gate capacitance of the second gate conductor layer 5b to which the word line WL is connected, by a combination of any of the lengths of the gate conductor layers 5a and 5b and the thicknesses and dielectric constants of the gate insulating layers 4a and 4b.

(28) FIGS. 2A to 2C are diagrams for explaining an effect attained in a case where the gate capacitance of the first gate conductor layer 5a connected to the plate line PL is made larger than the gate capacitance of the second gate conductor layer 5b to which the word line WL is connected.

(29) FIG. 2A is a simplified structural diagram of the dynamic flash memory cell according to the first embodiment of the present invention and illustrates only main parts. To the dynamic flash memory cell, the bit line BL, the word line WL, the plate line PL, and the source line SL are connected, and the potential state of the channel region 7 is determined by the voltage states of the lines.

(30) FIG. 2B is a diagram for explaining the capacitance relationships of the respective lines. The capacitance C.sub.FB of the channel region 7 is equal to the sum of the capacitance C.sub.WL between the gate conductor layer 5b to which the word line WL is connected and the channel region 7, the capacitance C.sub.PL between the gate conductor layer 5a to which the plate line PL is connected and the channel region 7, the junction capacitance C.sub.SL of the PN junction between the source N.sup.+ layer 3a to which the source line SL is connected and the channel region 7, and the junction capacitance C.sub.BL of the PN junction between the drain N.sup.+ layer 3b to which the bit line BL is connected and the channel region 7, and is expressed as follows.
C.sub.FB=C.sub.WL+C.sub.PL+C.sub.BL+C.sub.SL(1)

(31) Therefore, the coupling ratio .sub.WL between the word line WL and the channel region 7, the coupling ratio .sub.FL between the plate line PL and the channel region 7, the coupling ratio .sub.BL between the bit line BL and the channel region 7, and the coupling ratio .sub.SL between the source line SL and the channel region 7 are expressed as follows.
.sub.WL=C.sub.WL/(C.sub.WL+C.sub.PL+C.sub.BL+C.sub.SL)(2)
.sub.PL=C.sub.PL/(C.sub.WL+C.sub.PL+C.sub.BL+C.sub.SL)(3)
.sub.BL=C.sub.BL/(C.sub.WL+C.sub.PL+C.sub.BL+C.sub.SL)(4)
.sub.SL=C.sub.SL/(C.sub.WL+C.sub.PL+C.sub.BL+C.sub.SL)(5)

(32) Here, C.sub.PL>C.sub.WL holds, and therefore, this results in .sub.PL>.sub.WL.

(33) FIG. 2C is a diagram for explaining a change in the voltage V.sub.FB of the channel region 7 when the voltage V.sub.WL of the word line WL rises at the time of a read operation or a write operation and subsequently drops. Here, the potential difference V.sub.FB when the voltage V.sub.FB of the channel region 7 transitions from a low-voltage state V.sub.FBL to a high-voltage state V.sub.FBH in response to the voltage V.sub.WL of the word line WL rising from 0 V to a high-voltage state V.sub.WLH is expressed as follows.
V.sub.FB=V.sub.FBHV.sub.FBL=.sub.WLV.sub.WLH(6)

(34) The coupling ratio .sub.WL between the word line WL and the channel region 7 is small and the coupling ratio .sub.FL between the plate line PL and the channel region 7 is large, and therefore, V.sub.FB is small, and the voltage V.sub.FB of the channel region 7 negligibly changes even when the voltage V.sub.WL of the word line WL changes at the time of a read operation or a write operation.

(35) FIGS. 3AA to 3AC and FIG. 3B illustrate a memory write operation (which is an example of page write operation in the claims) of the dynamic flash memory cell according to the first embodiment of the present invention. FIG. 3AA illustrates a mechanism of the write operation, and FIG. 3AB illustrates operation waveforms of the bit line BL, the source line SL, the plate line PL, the word line WL, and the channel region 7 that functions as a floating body FB. At time TO, the dynamic flash memory cell is in a 0 erase state, and the voltage of the channel region 7 is equal to V.sub.FB0. Vss is applied to the bit line BL, the source line SL, and the word line WL, and V.sub.PLL is applied to the plate line PL. Here, for example, Vss is equal to 0 V and V.sub.PLL is equal to 2 V. Subsequently, from time T1 to time T2, when the bit line BL rises from Vss to V.sub.BLH, in a case where, for example, Vss is equal to 0 V, the voltage of the channel region 7 becomes equal to V.sub.FB0+.sub.BLV.sub.BLH due to capacitive coupling between the bit line BL and the channel region 7.

(36) The description of the write operation of the dynamic flash memory cell will be continued with reference to FIGS. 3AA and 3AB. From time T3 to time T4, the word line WL rises from Vss to V.sub.WLH. Accordingly, when the threshold voltage for 0 erase for a second N-channel MOS transistor region that is a region in which the second gate conductor layer 5b to which the word line WL is connected surrounds the channel region 7 is denoted by Vt.sub.WL0, as the voltage of the word line WL rises, in a range from Vss to Vt.sub.WL0, the voltage of the channel region 7 becomes equal to V.sub.FB0+.sub.BLV.sub.BLH+.sub.WLVt.sub.WL0 due to second capacitive coupling between the word line WL and the channel region 7. When the voltage of the word line WL rises to Vt.sub.WL0 or above, an inversion layer 12b in a ring form is formed in the channel region 7 on the inner periphery of the second gate conductor layer 5b and interrupts the second capacitive coupling between the word line WL and the channel region 7.

(37) The description of the write operation of the dynamic flash memory cell will be continued with reference to FIGS. 3AA and 3AB. From time T3 to time T4, for example, a fixed voltage V.sub.PLL=2 V is applied to the first gate conductor layer 5a to which the plate line PL is connected, and the second gate conductor layer 5b to which the word line WL is connected is increased to, for example, V.sub.WLH=4 V. As a result, as illustrated in FIG. 3AA, an inversion layer 12a in a ring form is formed in the channel region 7 on the inner periphery of the first gate conductor layer 5a to which the plate line PL is connected, and a pinch-off point 13 is present in the inversion layer 12a. As a result, a first N-channel MOS transistor region including the first gate conductor layer 5a operates in the saturation region. In contrast, the second N-channel MOS transistor region including the second gate conductor layer 5b to which the word line WL is connected operates in the linear region. As a result, a pinch-off point is not present in the channel region 7 on the inner periphery of the second gate conductor layer 5b to which the word line WL is connected, and the inversion layer 12b is formed on the entire inner periphery of the gate conductor layer 5b. The inversion layer 12b that is formed on the entire inner periphery of the second gate conductor layer 5b to which the word line WL is connected substantially functions as the drain of the second N-channel MOS transistor region including the second gate conductor layer 5b. As a result, the electric field becomes maximum in a first boundary region of the channel region 7 between the first N-channel MOS transistor region including the first gate conductor layer 5a and the second N-channel MOS transistor region including the second gate conductor layer 5b that are connected in series, and an impact ionization phenomenon occurs in this region. This region is a source-side region when viewed from the second N-channel MOS transistor region including the second gate conductor layer 5b to which the word line WL is connected, and therefore, this phenomenon is called a source-side impact ionization phenomenon. By this source-side impact ionization phenomenon, electrons flow from the N.sup.+ layer 3a to which the source line SL is connected toward the N.sup.+ layer 3b to which the bit line is connected. The accelerated electrons collide with lattice Si atoms, and electron-positive hole pairs are generated by the kinetic energy. Although some of the generated electrons flow into the first gate conductor layer 5a and into the second gate conductor layer 5b, most of the generated electrons flow into the N.sup.+ layer 3b to which the bit line BL is connected (not illustrated).

(38) As illustrated in FIG. 3AC, a generated group of positive holes 9 (which is an example of group of positive holes in the claims) are majority carriers in the channel region 7, with which the channel region 7 is charged to a positive bias. The N.sup.+ layer 3a to which the source line SL is connected is at 0 V, and therefore, the channel region 7 is charged up to the built-in voltage Vb (about 0.7 V) of the PN junction between the N.sup.+ layer 3a to which the source line SL is connected and the channel region 7. When the channel region 7 is charged to a positive bias, the threshold voltages for the first N-channel MOS transistor region and the second N-channel MOS transistor region decrease due to a substrate bias effect.

(39) The description of the write operation of the dynamic flash memory cell will be continued with reference to FIG. 3AB. From time T6 to time T7, the voltage of the word line WL drops from V.sub.WLH to Vss. During this period, although the second capacitive coupling is formed between the word line WL and the channel region 7, the inversion layer 12b interrupts the second capacitive coupling until the voltage of the word line WL drops from V.sub.WLH to a threshold voltage Vt.sub.WL1 for the second N-channel MOS transistor region or below when the voltage of the channel region 7 is equal to Vb. Therefore, the capacitive coupling between the word line WL and the channel region 7 is substantially formed only during a period from when the word line WL drops to Vt.sub.W L1 or below to when the word line WL drops to Vss. As a result, the voltage of the channel region 7 becomes equal to Vb.sub.WLVt.sub.WL1. Here, Vt.sub.WL1 is lower than Vt.sub.WL0 described above, and .sub.WLVt.sub.WL1 is small.

(40) The description of the write operation of the dynamic flash memory cell will be continued with reference to FIG. 3AB. From time T8 to time T9, the bit line BL drops from V.sub.BLH to Vss. The bit line BL and the channel region 7 are capacitively coupled with each other, and therefore, the 1 write voltage V.sub.FB1 of the channel region 7 becomes as follows at the end.
V.sub.FB1=Vb.sub.WLVt.sub.WL1.sub.BLV.sub.BLH(7)

(41) Here, the coupling ratio .sub.BL between the bit line BL and the channel region 7 is also small. Accordingly, as illustrated in FIG. 3B, the threshold voltage for the second N-channel MOS transistor region of the second channel region 7b to which the word line WL is connected decreases. The memory write operation in which the voltage V.sub.FB1 in the 1 write state of the channel region 7 is assumed to be a first data retention voltage (which is an example of first data retention voltage in the claims) is performed to assign logical storage data 1.

(42) At the time of the write operation, electron-positive hole pairs may be generated by an impact ionization phenomenon in a second boundary region between the first impurity layer 3a and the first channel semiconductor layer 7a or in a third boundary region between the second impurity layer 3b and the second channel semiconductor layer 7b instead of the first boundary region, and the channel region 7 may be charged with the generated group of positive holes 9.

(43) Note that the conditions of voltages applied to the bit line BL, the source line SL, the word line WL, and the plate line PL and the potential of the floating body described above are examples for performing the write operation, and other operation conditions based on which the write operation can be performed may be employed.

(44) A mechanism of a memory erase operation (which is an example of page erase operation in the claims) will be described with reference to FIG. 4A to FIGS. 4EA to 4ED.

(45) FIG. 4A is a memory block circuit diagram for explaining a page erase operation. Although nine memory cells CL.sub.11 to CL.sub.33 in three rows and three columns are illustrated, the actual memory block is larger than this matrix. When memory cells are arranged in a matrix, one of the directions of the arrangement is called a row direction (or in rows) and the direction perpendicular to the one of the directions is called column direction (or in columns). To each of the memory cells, the source line SL, a corresponding one of the bit lines BL.sub.1 to BL.sub.3, a corresponding one of the plate lines PL.sub.1 to PL.sub.3, and a corresponding one of the word lines WL.sub.1 to WL.sub.3 are connected. For example, it is assumed that the memory cells CL.sub.21 to CL.sub.23, in a certain page (which is an example of page in the claims), to which the plate line PL.sub.2 and the word line WL.sub.2 are connected are selected in this block and a page erase operation is performed.

(46) A mechanism of the page erase operation will be described with reference to FIGS. 4BA to 4BD and FIG. 4C. Here, the channel region 7 between the N.sup.+ layers 3a and 3b is electrically isolated from the substrate and functions as a floating body. FIG. 4BA is a timing operation waveform diagram of main nodes in the erase operation. In FIG. 4BA, T0 to T12 indicate times from the start to the end of the erase operation. FIG. 4BB illustrates a state at time T0 before the erase operation, in which the group of positive holes 9 generated by an impact ionization phenomenon in the previous cycle are stored in the channel region 7. From time T1 to time T2, the bit lines BL.sub.1 to BL.sub.3 and the source line SL rise from Vss to V.sub.BLH and V.sub.SLH respectively and are in a high-voltage state. Here, Vss is, for example, equal to 0 V. With this operation, during the subsequent period from time T3 to time T4, the plate line PL.sub.2 selected in the page erase operation rises from a first voltage V.sub.PLL to a second voltage V.sub.PLH and is in a high-voltage state, the word line WL.sub.2 selected in the page erase operation rises from a third voltage Vss to a fourth voltage V.sub.WLH and is in a high-voltage state, and this prevents the inversion layer 12a on the inner periphery of the first gate conductor layer 5a to which the plate line PL.sub.2 is connected and the inversion layer 12b on the inner periphery of the second gate conductor layer 5b to which the word line WL.sub.2 is connected from being formed in the channel region 7. Therefore, when the threshold voltage for the second N-channel MOS transistor region on the side of the word line WL.sub.2 and the threshold voltage for the first N-channel MOS transistor region on the side of the plate line PL.sub.2 are denoted by V.sub.tWL and V.sub.tPL respectively, it is desirable that the voltages V.sub.BLH and V.sub.SLH satisfy V.sub.BLH>V.sub.WLH+V.sub.tWL and V.sub.SLH>V.sub.PLH+V.sub.tPL. For example, in a case where V.sub.tWL and V.sub.tPL are equal to 0.5 V, V.sub.WLH and V.sub.PLH need to be set to 3 V, and V.sub.BLH and V.sub.SLH need to be set to 3.5 V or higher.

(47) The description of the mechanism of the page erase operation illustrated in FIG. 4BA will be continued. As the plate line PL.sub.2 and the word line WL respectively rise to the second voltage V.sub.PLH and the fourth voltage V.sub.WLH and are in a high-voltage state during a first period from time T3 to time T4, the voltage of the channel region 7 in a floating state is increased due to first capacitive coupling between the plate line PL.sub.2 and the channel region 7 and second capacitive coupling between the word line WL.sub.2 and the channel region 7. The voltage of the channel region 7 rises from V.sub.FB1 in the 1 write state to a high voltage. This voltage rise is possible because the voltage of the bit lines BL.sub.1 to BL.sub.3 and that of the source line SL are high voltages of V.sub.BLH and V.sub.SLH respectively and the PN junction between the source N.sup.+ layer 3a and the channel region 7 and the PN junction between the drain N.sup.+ layer 3b and the channel region 7 are in a reverse bias state accordingly.

(48) The description of the mechanism of the page erase operation illustrated in FIG. 4BA will be continued. During the subsequent period from time T5 to time T6, the voltage of the bit lines BL.sub.1 to BL.sub.3 and that of the source line SL respectively drop from high voltages of V.sub.BLH and V.sub.SLH to Vss. As a result, the PN junction between the source N.sup.+ layer 3a and the channel region 7 and the PN junction between the drain N.sup.+ layer 3b and the channel region 7 are in a forward bias state as illustrated in FIG. 4BC, and a remaining group of positive holes among the group of positive holes 9 in the channel region 7 are discharged to the source N.sup.+ layer 3a and to the drain N.sup.+ layer 3b. As a result, the voltage V.sub.FB of the channel region 7 becomes equal to the built-in voltage Vb of the PN junction formed by the source N.sup.+ layer 3a and the P-layer channel region 7 and the PN junction formed by the drain N.sup.+ layer 3b and the P-layer channel region 7.

(49) The description of the mechanism of the page erase operation illustrated in FIG. 4BA will be continued. Subsequently, from time T7 to time T8, the voltage of the bit lines BL.sub.1 to BL.sub.3 and that of the source line SL rise from Vss to high voltages of V.sub.BLH and V.sub.SLH respectively. With this operation, as illustrated in FIG. 4BD, when the plate line PL.sub.2 drops from the second voltage V.sub.PLH to the first voltage V.sub.PLL and the word line WL.sub.2 drops from the fourth voltage V.sub.WLH to the third voltage Vss, from time T9 to time T10, the voltage V.sub.FB of the channel region 7 efficiently changes from Vb to V.sub.FB0 due to the first capacitive coupling between the plate line PL.sub.2 and the channel region 7 and the second capacitive coupling between the word line WL.sub.2 and the channel region 7 without the inversion layer 12a on the side of the plate line PL.sub.2 or the inversion layer 12b on the side of the word line WL.sub.2 being formed in the channel region 7. The voltage difference V.sub.FB of the channel region 7 between the 1 write state and the 0 erase state is expressed by the following expressions.

(50) $\begin{array}{cc}{V}_{\mathrm{FB}}1=\mathrm{Vb}-{}_{WL}V{t}_{WL}1-{}_{BL}{V}_{\mathrm{BLH}}& \left(7\right)\end{array}$$\begin{array}{cc}{V}_{FB}0=\mathrm{Vb}-{}_{WL}{V}_{\mathrm{WLH}}-{}_{PL}\left(V_{\mathrm{PLH}}-V_{\mathrm{PLL}}\right)& \left(8\right)\end{array}$$\begin{array}{cc}{V}_{FB}=V_{FB}1-V_{\mathrm{FB}}0={}_{WL}{V}_{\mathrm{WLH}}+{}_{PL}\left(V_{\mathrm{PLH}}-V_{\mathrm{PLL}}\right)-{}_{WL}V{t}_{WL}1-{}_{BL}{V}_{\mathrm{BLH}}& \left(9\right)\end{array}$

(51) Here, the sum of .sub.WL and .sub.PL is greater than or equal to 0.8, V.sub.FB is large, and a sufficient margin is provided.

(52) As a result, as illustrated in FIG. 4C, a large margin is provided between the 1 write state and the 0 erase state. Here, in the 0 erase state, the threshold voltage on the side of the plate line PL.sub.2 is high due to a substrate bias effect. Therefore, when the voltage applied to the plate line PL.sub.2 is set to, for example, the threshold voltage or lower, the first N-channel MOS transistor region on the side of the plate line PL.sub.2 becomes non-conducting and does not allow the memory cell current to flow therethrough. This state is illustrated in the right part of FIG. 4C and indicated as PL: non-conducting.

(53) The description of the mechanism of the page erase operation illustrated in FIG. 4BA will be continued. During the subsequent period, namely, a fourth period from time T11 to time T12, the voltage of the bit lines BL.sub.1 to BL.sub.3 drops from V.sub.BLH to Vss and that of the source line SL drops from V.sub.SLH to Vss, and the erase operation ends. At this time, although the bit lines BL.sub.1 to BL.sub.3 and the source line SL slightly decrease the voltage of the channel region 7 due to capacitive coupling, this decrease is equal to the increase in the voltage of the channel region 7 by the bit lines BL.sub.1 to BL.sub.3 and the source line SL from time T7 to time T8 due to capacitive coupling, and therefore, the decrease and the increase in the voltage by the bit lines BL.sub.1 to BL.sub.3 and the source line SL are canceled out, and the voltage of the channel region 7 is not affected consequently. The page erase operation in which the voltage V.sub.FB0 in the 0 erase state of the channel region 7 is assumed to be a second data retention voltage (which is an example of second data retention voltage in the claims) is performed to assign logical storage data 0.

(54) Now, a mechanism of the page erase operation will be described with reference to FIGS. 4DA to 4DD. FIGS. 4DA to 4DD are different from FIGS. 4BA to 4BD in that the bit lines BL.sub.1 to BL.sub.3 are kept at Vss or put in a floating state and the word line WL.sub.2 is kept at Vss during the page erase operation. Accordingly, from time T1 to time T2, even when the source line SL rises from Vss to V.sub.SLH, the second N-channel MOS transistor region of the word line WL.sub.2 is non-conducting, and the memory cell current does not flow therethrough. Therefore, the group of positive holes 9 caused by an impact ionization phenomenon are not generated. The others are the same as in FIGS. 4BA to 4BD, and the source line SL changes between Vss and V.sub.SLH, and the plate line PL.sub.2 changes between V.sub.PLL and V.sub.PLH. As a result, as illustrated in FIG. 4DC, the group of positive holes 9 are discharged to the first impurity layer, namely, the N.sup.+ layer 3a, of the source line SL.

(55) Now, a mechanism of the page erase operation will be described with reference to FIGS. 4EA to FIG. 4ED. FIGS. 4EA to FIG. 4ED are different from FIGS. 4BA to 4BD in that the source line SL is kept at Vss or put in a floating state and the plate line PL.sub.2 is kept at Vss during the page erase operation. Accordingly, from time T1 to time T2, even when the bit lines BL.sub.1 to BL.sub.3 rise from Vss to V.sub.BLH, the first N-channel MOS transistor region of the plate line PL.sub.2 is non-conducting, and the memory cell current does not flow therethrough. Therefore, the group of positive holes 9 caused by an impact ionization phenomenon are not generated. The others are the same as in FIGS. 4BA to 4BD, and the bit lines BL.sub.1 to BL.sub.3 change between Vss and V.sub.BLH, and the word line WL.sub.2 changes between Vss and V.sub.WLH. As a result, as illustrated in FIG. 4EC, the group of positive holes 9 are discharged to the second impurity layer, namely, the N.sup.+ layer 3b, of the bit lines BL.sub.1 to BL.sub.3.

(56) Note that the conditions of voltages applied to the bit line BL, the source line SL, the word line WL, and the plate line PL and the potential of the floating body described above are examples for performing the page erase operation, and other operation conditions based on which the page erase operation can be performed may be employed.

(57) FIGS. 5A to 5C are diagrams for explaining a read operation of the dynamic flash memory cell according to the first embodiment of the present invention. As illustrated in FIG. 5A, when the channel region 7 is charged up to the built-in voltage Vb (about 0.7 V), the threshold voltage for the second N-channel MOS transistor region including the second gate conductor layer 5b to which the word line WL is connected decreases due to a substrate bias effect. This state is assigned to logical storage data 1. As illustrated in FIG. 5B, a memory block selected before writing is in an erase state 0 in advance, and the voltage V.sub.FB of the channel region 7 is equal to V.sub.FB0. With a write operation, a write state 1 is stored at random. As a result, logical storage data of logical 0 and that of logical 1 are created for the word line WL. As illustrated in FIG. 5C, the level difference between the two threshold voltages of the word line WL is used to perform reading by a sense amplifier. When the voltage applied to the first gate conductor layer 5a connected to the plate line PL is set to a voltage higher than the threshold voltage at the time of logical storage data 1 and lower than the threshold voltage at the time of logical storage data 0 in logical 0 data reading, a property that a current does not flow even when the voltage of the word line WL is increased can be attained as illustrated in FIG. 5C.

(58) Note that the conditions of voltages applied to the bit line BL, the source line SL, the word line WL, and the plate line PL and the potential of the floating body described above are examples for performing the read operation, and other operation conditions based on which the read operation can be performed may be employed.

(59) FIGS. 6AA and 6AB and FIG. 6B are diagrams for explaining a first refresh operation (which is an example of first refresh operation in the claims) for a 0 erase state of the dynamic flash memory cell according to the first embodiment of the present invention.

(60) FIG. 6AA illustrates a single dynamic flash memory cell in a 0 erase state. Although the initial voltage of the channel region 7 to which 0 is written is equal to V.sub.FB0 at a first time (which is an example of first time in the claims), the voltage changes over time and rises to V.sub.FB0+V at a second time (which is an example of second time in the claims). This is caused by a leak current of, for example, the group of positive holes 9 flowing into the channel region 7. FIG. 6AB illustrates four dynamic flash memory cells CL00 to CL03 that constitute a part of the same page. While the dynamic flash memory cells CL00 and CL03 are in a 1 write state and an excessive group of positive holes 9 are retained in the channel region 7, the dynamic flash memory cells CL01 and CL02 are in a 0 erase state, and no excessive group of positive holes 9 are retained in the channel region 7 but a small group of positive holes 9 have entered the channel region 7 due to, for example, a leak current. To perform the first refresh operation, which is a memory re-erase operation, of the dynamic flash memory cells CL01 and CL02 in the 0 erase state, a positive-bias pulse is applied to the word line WL and to the plate line PL while the bit lines BL.sub.0 to BL.sub.3 and the source line SL are kept at Vss. Here, Vss is equal to, for example, 0 V. Specific examples of the operation waveforms are illustrated in FIG. 6B, which illustrates changes in the voltages of the bit line BL, the source line SL, the word line WL, the plate line PL, and the channel region 7 that functions as the floating body FB, over time from time T0 to time T2. At time T0, the voltage of the channel region 7 of the dynamic flash memory cell in the 0 erase state rises to V.sub.FB3=V.sub.FB0+V.

(61) When the word line WL rises from Vss to V.sub.WLRE and the plate line PL rises from V.sub.PLL to V.sub.PLR at time T1, due to second capacitive coupling between the second gate conductor layer 5b to which the word line WL is connected and the channel layer 7 and first capacitive coupling between the first gate conductor layer 5a to which the plate line PL is connected and the channel layer 7, the voltage of the channel layer 7 becomes equal to V.sub.FB4 that is higher than the built-in voltage Vb. As a result, in the dynamic flash memory cells CL01 and CL02 in the 0 erase state, the PN junction between the P-layer channel region 7 and the source N.sup.+ layer 3a and the PN junction between the P-layer channel region 7 and the drain N.sup.+ layer 3b are forward biased, and the group of positive holes 9 having flowed into the channel region 7 are discharged to the source N.sup.+ layer 3a and to the drain N.sup.+ layer 3b. In the dynamic flash memory cells CL00 and CL03 in the 1 write state, the inversion layer 12a is formed in the channel region 7 surrounded by the first gate conductor layer 5a to which the plate line PL is connected and the inversion layer 12b is formed in the channel region 7 surrounded by the second gate conductor layer 5b to which the word line WL is connected. Both the bit line BL and the source line SL are kept at Vss, and therefore, a current does not flow from the bit line BL to the source line SL or an impact ionization phenomenon does not occur. Further, the inversion layers 12a and 12b shield the first gate conductor layer 5a and the second gate conductor layer 5b from the channel region 7, and therefore, the voltage of the channel region 7 does not rise.

(62) The description of the first refresh operation for the 0 erase state of the dynamic flash memory cell according to the first embodiment of the present invention will be continued with reference to FIG. 6B. When the voltage of the word line WL drops from V.sub.WLRE to Vss and the voltage of the plate line PL drops from V.sub.PLR to V.sub.PLL at time T2, the voltage V.sub.FB0 of the channel region 7 of the dynamic flash memory cell in the 0 erase state after the refresh operation is expressed by expression (10).
V.sub.FB0=Vb.sub.WLV.sub.WLRE.sub.PL(V.sub.PLRV.sub.PLL)(10)

(63) Therefore, the threshold voltage Vt.sub.WL0 for the second N-channel MOS transistor region that is a region in which the second gate conductor layer 5b surrounds the channel layer 7 increases from that in the 0 erase state at the second time before the first refresh operation.

(64) FIGS. 6CA and 6CB and FIG. 6D are diagrams for explaining a second refresh operation (which is an example of second refresh operation in the claims) for a 1 write state of the dynamic flash memory cell according to the first embodiment of the present invention.

(65) FIG. 6CA illustrates a single dynamic flash memory cell constituted by a single semiconductor base material and in a 1 write state. Although the initial voltage of the channel region 7 to which 1 is written is equal to V.sub.FB1, the voltage changes over time and drops to V.sub.FB1V. This is caused by a leak current of the group of positive holes 9 from the channel region 7, that is, for example, a leak current to the source N.sup.+ layer 3a and to the drain N.sup.+ layer 3b. FIG. 6CB illustrates the four dynamic flash memory cells CL00 to CL03 that constitute a part of the same page in which a plurality of semiconductor base materials are arranged. Here, the dynamic flash memory cells CL00 and CL03 are in a 1 write state at the first time and an excessive group of positive holes 9 are retained in the channel region 7. The dynamic flash memory cells CL01 and CL02 are in a 0 erase state and no excessive group of positive holes 9 are retained in the channel region 7. To perform the second refresh operation, which is a memory re-write operation, of the dynamic flash memory cells CL00 and CL03 in the 1 write state, a positive bias is applied to the bit lines BL.sub.0 to BL.sub.3, to the word line WL, and to the plate line PL and the source line SL is kept at Vss. Here, Vss is equal to, for example, 0 V. Specific examples of the operation waveforms are illustrated in FIG. 6D, which illustrates changes in the voltages of the bit line BL, the source line SL, the word line WL, the plate line PL, and the channel region 7 that functions as the floating body FB, over time from time T0 to time T4. At time T0 that is a third time (which is an example of third time in the claims), the voltage of the channel region 7 of the dynamic flash memory cell in the 1 write state drops to V.sub.FB1=V.sub.FB1V. At time T1, the voltage of the bit line BL rises from Vss to V.sub.BLR. Here, V.sub.BLR is a voltage of the bit line BL for a refresh operation of a block to which 1 is written and is equal to, for example, 2 V. As a result, due to capacitive coupling of the PN junction between the drain N.sup.+ layer 3b and the P-layer channel region 7, when the coupling ratio of capacitive coupling between the bit line BL and the channel region 7 is denoted by .sub.BL, the voltage V.sub.FB2 of the channel region 7 is expressed by expression (11) by using expression (4).
.sub.BL=C.sub.BL/(C.sub.WL+C.sub.PL+C.sub.BL+C.sub.SL)(4)
V.sub.FB2=V.sub.FB1+.sub.BLV.sub.BLR(11)

(66) The description of the second refresh operation for the 1 write state of the dynamic flash memory cell according to the first embodiment of the present invention will be continued with reference to FIG. 6D. At time T2, the word line WL rises from Vss to a high voltage of V.sub.WLRP. Here, V.sub.WLRP is a voltage applied to the word line WL for a refresh operation of a block to which 1 is written and is equal to, for example, 4 V. As a result, in the dynamic flash memory cells CL00 and CL03 in the 1 write state, the first N-channel MOS transistor region that is a region in which the first gate conductor layer 5a to which the plate line PL is connected surrounds the channel region 7 operates in the saturation region, and the second N-channel MOS transistor region that is a region in which the second gate conductor layer 5b to which the word line WL is connected surrounds the channel layer 7 operates in the linear region. Accordingly, a current flows from the drain N.sup.+ layer 3b to the source N.sup.+ layer 3a, and the group of positive holes 9 are generated by an impact ionization phenomenon, with which the channel region 7 is filled. As a result, the voltage of the channel region 7 becomes equal to the built-in voltage Vb (about 0.7 V). In contrast, in the dynamic flash memory cells CL01 and CL02 that are in the 0 erase state, the threshold voltage Vt.sub.PL0 in the 0 erase state of the first N-channel MOS transistor region that is a region in which the first gate conductor layer 5a to which the plate line PL is connected surrounds the channel layer 7 and the threshold voltage Vt.sub.WL0 in the 0 erase state of the second N-channel MOS transistor region that is a region in which the second gate conductor layer 5b to which the word line WL is connected surrounds the channel layer 7 are high, and therefore, a current does not flow from the drain N.sup.+ layer 3b to the source N.sup.+ layer 3a at time T2 or an impact ionization phenomenon does not occur.

(67) The description of the second refresh operation for the 1 write state of the dynamic flash memory cell according to the first embodiment of the present invention will be continued with reference to FIG. 6D. At time T3, the word line WL drops from V.sub.WLRP to Vss. The inversion layer 12b is generated in the channel layer 7 surrounded by the second gate conductor layer 5b to which the word line WL is connected, and this inversion layer interrupts second capacitive coupling between the second gate conductor layer 5b and the channel layer 7. When the threshold voltage in the 1 write state of the second N-channel MOS transistor region that is a region in which the second gate conductor layer 5b surrounds the channel layer 7 is denoted by Vt.sub.WL1, the second capacitive coupling between the second gate conductor layer 5b and the channel layer 7 works when the voltage of the word line WL drops to Vt.sub.WL1 or below, and the voltage of the channel region 7 becomes equal to Vb1 expressed by expression (12).
Vb1=Vb.sub.WLVt.sub.WL1(12)

(68) Here, the threshold voltage Vt.sub.WL1 in the 1 write state is low, and therefore, .sub.WLVt.sub.WL1 is small. Subsequently, at time T4, the bit line BL drops from V.sub.BLR to Vss, and the second refresh operation for the 1 write state ends. The voltage V.sub.FB1 of the channel region 7 becomes equal to Vb2 at the end expressed by expression (13).
V.sub.FB1=Vb2=Vb1.sub.BLV.sub.BLR=Vb.sub.WLVt.sub.WL1.sub.BLV.sub.BLR(13)

(69) Therefore, the threshold voltage Vt.sub.WL for the second N-channel MOS transistor region that is a region in which the second gate conductor layer 5b surrounds the channel layer 7 decreases from that in the 1 write state before the second refresh operation.

(70) In FIG. 6E, memory cells C00 to C22 in three rows and three columns constitute a part of a block (which is an example of block in the claims). Although the memory cells C00 to C22 in three rows and three columns are illustrated here, memory cells in a matrix larger than three rows and three columns constitute the actual block. To each of the memory cells, a corresponding one of the word lines WL0 to WL2, a corresponding one of the plate lines PL0 to PL2, the source line SL, and a corresponding one of the bit lines BL0 to BL2 are connected. Transistors T0C to T2C each having a gate to which a transfer signal FT is input constitute a switch circuit. The drains of transistors T0D to T2D each having a gate connected to a bit line pre-charge signal FS are connected to a bit line power VB, and the sources thereof are connected to the bit lines BL0 to BL2 respectively. The bit lines BL0 to BL2 are respectively connected to sense amplifier circuits SA0 to SA2 with the switch circuit therebetween. The word lines WL0 to WL2 and the plate lines PL0 to PL2 are connected to a row decoder circuit RDEC. The sense amplifier circuits SA0 to SA2 are connected to a pair of complementary input/output lines IO and/IO, with transistors T0A to T2B each having a gate connected to a corresponding one of the column selection lines CSL0 to CSL2 therebetween.

(71) FIG. 6F is a circuit block diagram illustrating a state in which, at a certain timing, 1 is written to the memory cells C10, C01, C21, C02, and C12 at random among the memory cells C00 to C22 and the group of positive holes 9 are stored in the channel semiconductor layer 7 of each of the memory cells. The first refresh operation related to all word lines WL and all plate lines PL in the block will be described with reference to FIG. 6G. FIG. 6G illustrates a case where an all-word-line selection signal ALLWL (which is an example of all-word-line selection signal in the claims) and an all-plate-line selection signal ALLPL (which is an example of all-plate-line selection signal in the claims) are input to the row decoder circuit. A state in which all the word lines WL0 to WL2 in the memory cell block are selected and the memory cells C10, C01, C21, C02, and C12 are subjected to the first refresh operation is illustrated. Although the memory cells C00, C11, C20, and C22 in FIG. 6G are in a 0 erase state and no excessive group of positive holes 9 are retained in the channel region 7, a small group of positive holes 9 entered the channel region 7 at the second time due to, for example, a leak current. To perform the first refresh operation, which is a memory re-erase operation, for the memory cells C00, C11, C20, and C22 in the 0 erase state, a positive-bias pulse is applied to the word lines WL and to the plate lines PL while the bit lines BL0 to BL2 and the source line SL are kept at Vss. Here, Vss is equal to, for example, 0 V. As a result, the group of positive holes 9 having entered the channel region 7 due to, for example, a leak current are discharged to the bit lines BL0 to BL2 and to the source line SL. The voltage of the channel region 7 of each of the memory cells C00, C11, C20, and C22 in the 0 erase state returns to a voltage close to the second data retention voltage.

(72) The second refresh operation related to all word lines WL and all plate lines PL in the block will be described with reference to FIG. 6H. A refresh operation of generating by an impact ionization phenomenon, the group of positive holes 9 inside the channel semiconductor layer 7 of each of the memory cells C10, C01, C21, C02, and C12 to which 1 is written in FIG. 6H is performed. The threshold voltage of the first gate conductor layer 5a and that of the second gate conductor layer 5b of each of the memory cells C10, C01, C21, C02, and C12 decrease, for example, from 1.3 V in the 0 erase state to 0.3 V in the 1 write state, that is, decrease by 1 V. Therefore, when a voltage lower than that in the 1 write operation is input to the plate lines PL0 to PL2 and to the word lines WL0 to WL2, the refresh operation of generating by an impact ionization phenomenon, the group of positive holes 9 inside the channel semiconductor layer 7 of each of the memory cells C10, C01, C21, C02, and C12 can be performed. Here, a case is assumed where the voltage of the channel semiconductor layer 7 of each of the memory cells C10, C01, C21, C02, and C12 to which 1 is written in FIG. 6H slightly decreases from the first data retention voltage V.sub.FB1 to V.sub.FB1-V.sub.FB.

(73) As described above, the first refresh operation and the second refresh operation related to all word lines WL and all plate lines PL in the memory cell block illustrated in FIG. 6G and FIG. 6H are very effective for a page sum-of-products read operation (which is an example of page sum-of-products read operation in the claims) in which at least two word lines are selected in multiple selection. In the page sum-of-products read operation, for example, the three word lines WL0 to WL2 in FIG. 6F are selected in multiple selection. Memory cell currents Icell are added up for each of the bit lines BL0 to BL2, and therefore, as the memory cell currents, saturated and stable memory cell currents need to be read. Therefore, the first refresh operation and the second refresh operation related to all word lines WL and all plate lines PL in the memory cell block in FIG. 6G and FIG. 6H performed before the page sum-of-products read operation are very effective.

(74) FIG. 6I illustrates an example where the plate line PL is shared among adjacent memory cells in the block constituted by the memory cells C00 to C22 in three rows and three columns. In this configuration, the first refresh operation and the second refresh operation of the dynamic flash memory cell according to the first embodiment of the present invention can also be performed before page reading.

(75) FIG. 6J illustrates an example where isolated source lines, namely, a source line SL0 parallel to the plate line PL0 and the word line WL0, a source line SL1 parallel to the plate line PL1 and the word line WL1, and a source line SL2 parallel to the plate line PL2 and the word line WL2, are disposed in the block constituted by the memory cells C00 to C22 in three rows and three columns. In this configuration, the first refresh operation and the second refresh operation of the dynamic flash memory cell according to the first embodiment of the present invention can also be performed before page reading.

(76) Regardless of whether the horizontal cross-sectional shape of the Si pillar 2 illustrated in FIG. 1 is a round shape, an elliptic shape, or a rectangular shape, the operations of the dynamic flash memory described in this embodiment can be performed. Further, a dynamic flash memory cell having a round shape, a dynamic flash memory cell having an elliptic shape, and a dynamic flash memory cell having a rectangular shape may coexist on the same chip.

(77) With reference to FIG. 1, the dynamic flash memory element including, for example, an SGT in which the first gate insulating layer 4a and the second gate insulating layer 4b that surround the entire side surface of the Si pillar 2 standing on the substrate in the vertical direction are provided and which includes the first gate conductor layer 5a and the second gate conductor layer 5b that entirely surround the first gate insulating layer 4a and the second gate insulating layer 4b has been described. As indicated in the description of this embodiment, the dynamic flash memory element needs to have a structure that satisfies the condition that the group of positive holes 9 generated by an impact ionization phenomenon are retained in the channel region 7. For this, the channel region 7 needs to have a floating body structure isolated from the substrate. Accordingly, even when the semiconductor base material of the channel region is formed horizontally along the substrate by using, for example, GAA (Gate All Around, see, for example, J. Y. Song, W. Y. Choi, J. H. Park, J. D. Lee, and B-G. Park: Design Optimization of Gate-All-Around (GAA) MOSFETs, IEEE Trans. Electron Devices, Vol. 5, No. 3, pp. 186-191, May 2006) technology, which is one type of SGT, or nanosheet technology (see, for example, N. Loubet, et al.: Stacked Nanosheet Gate-All-Around Transistor to Enable Scaling Beyond FinFET, 2017 IEEE Symposium on VLSI Technology Digest of Technical Papers, T17-5, T230-T231, June 2017), the above-described operations of the dynamic flash memory can be performed. Alternatively, the dynamic flash memory element may have a device structure using SOI (Silicon On Insulator) (see, for example, J. Wan, L. Rojer, A. Zaslaysky, and S. Critoloveanu: A Compact Capacitor-Less High-Speed DRAM Using Field Effect-Controlled Charge Regeneration, Electron Device Letters, Vol. 35, No. 2, pp. 179-181 (2012), T. Ohsawa, K. Fujita, T. Higashi, Y. Iwata, T. Kajiyama, Y. Asao, and K. Sunouchi: Memory design using a one-transistor gain cell on SOI, IEEE JSSC, Vol. 37, No. 11, pp. 1510-1522 (2002), T. Shino, N. Kusunoki, T. Higashi, T. Ohsawa, K. Fujita, K. Hatsuda, N. Ikumi, F. Matsuoka, Y. Kajitani, R. Fukuda, Y. Watanabe, Y. Minami, A. Sakamoto, J. Nishimura, H. Nakajima, M. Morikado, K. Inoh, T. Hamamoto, A. Nitayama: Floating Body RAM Technology and its Scalability to 32 nm Node and Beyond, IEEE IEDM (2006), and E. Yoshida: A Capacitorless 1T-DRAM Technology Using Gate-Induced Drain-Leakage (GIDL) Current for Low-Power and High-Speed Embedded Memory, IEEE IEDM (2006)). In this device structure, the bottom portion of the channel region is in contact with an insulating layer of the SOI substrate, and the other portion of the channel region is surrounded by a gate insulating layer and an element isolation insulating layer. With such a structure, the channel region also has a floating body structure. Accordingly, the dynamic flash memory element provided in this embodiment needs to satisfy the condition that the channel region has a floating body structure. Even with a structure in which a Fin transistor (see, for example, H. Jiang, N. Xu, B. Chen, L. Zeng 1, Y. He, G. Du, X. Liu and X. Zhang: Experimental investigation of self-heating effect (SHE) in multiple-fin SOI FinFETs, Semicond. Sci. Technol. 29 (2014) 115021 (7pp)) is formed on an SOI substrate, as long as the channel region has a floating body structure, the dynamic flash operations can be performed.

(78) Expressions (1) to (13) provided in the specification and in the drawings are expressions used to qualitatively explain the phenomena, and are not intended to limit the phenomena.

(79) Although the reset voltages of the word line WL, the bit line BL, and the source line SL are specified as Vss in the explanations of FIGS. 3AA to 3AC and FIG. 3B, the reset voltages of the respective lies may be set to different voltages.

(80) Although FIG. 4BA, FIG. 4DA, and FIG. 4EA and the descriptions thereof illustrate example conditions of the page erase operation, the voltages applied to the source line SL, the plate line PL, the bit line BL, and the word line WL may be changed as long as a state in which the group of positive holes 9 in the channel region 7 are discharged through either the N.sup.+ layer 3a or the N.sup.+ layer 3b or both the N.sup.+ layer 3a and the N.sup.+ layer 3b can be attained. Further, in the page erase operation, a voltage may be applied to the source line SL of a selected page, and the bit line BL may be put in a floating state. In the page erase operation, a voltage may be applied to the bit line BL of a selected page, and the source line SL may be put in a floating state.

(81) In FIG. 1, in the vertical direction, in a part of the channel region 7 surrounded by the insulating layer 6 that is the first insulating layer, the potential distribution of the first channel region 7a and that of the second channel region 7b are connected and formed. Accordingly, the first channel region 7a and the second channel region 7b that constitute the channel region 7 are connected in the vertical direction in the region surrounded by the insulating layer 6 that is the first insulating layer.

(82) Note that in FIG. 1, it is desirable to make the length of the first gate conductor layer 5a, in the vertical direction, to which the plate line PL is connected further longer than the length of the second gate conductor layer 5b, in the vertical direction, to which the word line WL is connected to attain C.sub.PL>C.sub.WL. However, when only the plate line PL is added, the coupling ratio (C.sub.WL/(C.sub.PL+C.sub.WL+C.sub.BL+C.sub.SL)), of capacitive coupling, of the word line WL to the channel region 7 decreases. As a result, the potential change V.sub.FB of the channel region 7 that is a floating body decreases.

(83) Although specific voltages of the bit line BL, the word line WL, and the plate line PL are indicated in the operation waveform diagram in FIG. 6D, any voltage conditions may be employed as long as an impact ionization phenomenon can be made to occur inside the channel region 7 with a memory cell current and the refresh operation of generating the group of positive holes 9 can be performed.

(84) As the voltage V.sub.PLL of the plate line PL, a fixed voltage of, for example, about 1 V may be applied.

(85) Note that in the specification and the claims, the meaning of cover in a case of for example, a gate insulating layer or a gate conductor layer covers, for example, a channel also includes a case of surrounding entirely as in an SGT or GAA, a case of surrounding except a portion as in a Fin transistor, and a case of overlapping a flat object as in a planar transistor.

(86) Although the first gate conductor layer 5a entirely surrounds the first gate insulating layer 4a in FIG. 1, a structure may be employed in which the first gate conductor layer 5a partially surrounds the first gate insulating layer 4a in plan view. The first gate conductor layer 5a may be divided into at least two gate conductor layers and each of which may be operated as a plate line PL electrode. Similarly, the second gate conductor layer 5b may be divided into two or more gate conductor layers, and the gate conductor layers may each function as a conductive electrode of the word line and may be operated synchronously or asynchronously. Accordingly, the operations of the dynamic flash memory can be performed.

(87) In FIG. 1, the first gate conductor layer 5a may be connected to the word line WL, and the second gate conductor layer 5b may be connected to the plate line PL. In this case, the operations of the dynamic flash memory described above can also be performed.

(88) Although the refresh operations of a one-bit dynamic flash memory cell constituted by a single semiconductor base material have been described with reference to FIGS. 6AA and 6AB to FIG. 6J, the present invention is also effective for the refresh operations of a one-bit high-speed dynamic flash memory cell constituted by two semiconductor base materials that store complementary data of 1 and 0.

(89) Although the refresh operations of a single-layer memory array including one-bit dynamic flash memory cells each constituted by a single semiconductor base material have been described with reference to FIGS. 6AA and 6AB to FIG. 6J, the present invention is also effective for a multilayer memory array including one-bit dynamic flash memory cells stacked in multiple tiers and each constituted by a single semiconductor base material.

(90) This embodiment has the following features.

(91) Feature 1

(92) The dynamic flash memory cell of this embodiment is constituted by the N.sup.+ layers 3a and 3b that function as the source and the drain, the channel region 7, the first gate insulating layer 4a, the second gate insulating layer 4b, the first gate conductor layer 5a, and the second gate conductor layer 5b, which are formed in a pillar form as a whole. The N.sup.+ layer 3a that functions as the source is connected to the source line SL, the N.sup.+ layer 3b that functions as the drain is connected to the bit line BL, the first gate conductor layer 5a is connected to the plate line PL, and the second gate conductor layer 5b is connected to the word line WL. A structure is employed in which the gate capacitance of the first gate conductor layer 5a to which the plate line PL is connected is larger than the gate capacitance of the second gate conductor layer 5b to which the word line WL is connected, which is a feature. In the dynamic flash memory cell, the first gate conductor layer and the second gate conductor layer are stacked in the vertical direction. Accordingly, even when the structure is employed in which the gate capacitance of the first gate conductor layer 5a to which the plate line PL is connected is larger than the gate capacitance of the second gate conductor layer 5b to which the word line WL is connected, the memory cell area does not increase in plan view. Accordingly, a high-performance and highly integrated dynamic flash memory cell can be implemented.

(93) Feature 2

(94) The first refresh operation of the dynamic flash memory cell according to the first embodiment of the present invention is performed prior to the second refresh operation. In the first refresh operation, the channel region 7 in a floating state is made to change to a large degree by using the word line WL and the plate line PL and the group of positive holes 9 having flowed into the channel region 7 due to, for example, a leak current are discharged to the source N.sup.+ layer 3a and to the drain N.sup.+ layer 3b. In a memory cell in a 1 write state, inversion layers are formed in the channel region 7 and the inversion layers interrupt capacitive coupling between the word line WL and the channel region 7 and between the plate line PL and the channel region 7. However, a very small number of positive holes in the channel region 7 in the 1 write state are discharged. In the second refresh operation, a current flows through only a memory cell in the 1 write state and the group of positive holes 9 caused by an impact ionization phenomenon are re-generated selectively in the channel region 7 of the memory cell in the 1 write state. Therefore, as described in the first embodiment of the present invention, it is very desirable to perform the second refresh operation of a memory cell in the 1 write state after the first refresh operation of a memory cell in the 0 erase state. When the second refresh operation is performed after the first refresh operation as in the present invention, a highly reliable memory device can be provided.

(95) Feature 3

(96) In terms of the roles of the first gate conductor layer 5a to which the plate line PL is connected in the dynamic flash memory cell according to the first embodiment of the present invention, in the write operation and in the read operation performed by the dynamic flash memory cell, the voltage of the word line WL changes. At this time, the plate line PL assumes the role of decreasing the capacitive coupling ratio between the word line WL and the channel region 7. As a result, an effect on changes in the voltage of the channel region 7 when the voltage of the word line WL changes can be substantially suppressed. Accordingly, the difference between the threshold voltages for the SGT transistors of the word line WL indicating logical 0 and logical 1 can be increased. This leads to an increased operation margin of the dynamic flash memory cell. When the voltage applied to the first gate conductor layer 5a connected to the plate line PL is set to a voltage higher than the threshold voltage at the time of logical storage data 1 and lower than the threshold voltage at the time of logical storage data 0 in logical 0 data reading, a property that a current does not flow even when the voltage of the word line WL is increased can be attained. This leads to a further increased operation margin of the dynamic flash memory cell.

Other Embodiments

(97) Although the Si pillar is formed in the present invention, the Si pillar may be a semiconductor pillar made of a semiconductor material other than Si. This is similarly applicable to other embodiments according to the present invention.

(98) Various embodiments and modifications can be made to the present invention without departing from the spirit and scope of the present invention in a broad sense. The above-described embodiments are intended to explain examples of the present invention and are not intended to limit the scope of the present invention. Any of the above-described embodiments and modifications can be combined. Further, the above-described embodiments from which some of the configuration requirements are removed as needed are also within the scope of the technical spirit of the present invention.

(99) With the semiconductor-element-including memory device according to the present invention, a high-density and high-performance dynamic flash memory that is an SGT-including memory device can be obtained.