WO2004107350A1 - 強誘電体メモリ - Google Patents
強誘電体メモリ Download PDFInfo
- Publication number
- WO2004107350A1 WO2004107350A1 PCT/JP2003/006601 JP0306601W WO2004107350A1 WO 2004107350 A1 WO2004107350 A1 WO 2004107350A1 JP 0306601 W JP0306601 W JP 0306601W WO 2004107350 A1 WO2004107350 A1 WO 2004107350A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- data
- ferroelectric
- voltage
- ferroelectric memory
- read
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Classifications
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11C—STATIC STORES
- G11C29/00—Checking stores for correct operation ; Subsequent repair; Testing stores during standby or offline operation
- G11C29/02—Detection or location of defective auxiliary circuits, e.g. defective refresh counters
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11C—STATIC STORES
- G11C11/00—Digital stores characterised by the use of particular electric or magnetic storage elements; Storage elements therefor
- G11C11/21—Digital stores characterised by the use of particular electric or magnetic storage elements; Storage elements therefor using electric elements
- G11C11/22—Digital stores characterised by the use of particular electric or magnetic storage elements; Storage elements therefor using electric elements using ferroelectric elements
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11C—STATIC STORES
- G11C29/00—Checking stores for correct operation ; Subsequent repair; Testing stores during standby or offline operation
- G11C29/02—Detection or location of defective auxiliary circuits, e.g. defective refresh counters
- G11C29/023—Detection or location of defective auxiliary circuits, e.g. defective refresh counters in clock generator or timing circuitry
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11C—STATIC STORES
- G11C29/00—Checking stores for correct operation ; Subsequent repair; Testing stores during standby or offline operation
- G11C29/02—Detection or location of defective auxiliary circuits, e.g. defective refresh counters
- G11C29/028—Detection or location of defective auxiliary circuits, e.g. defective refresh counters with adaption or trimming of parameters
Definitions
- the present invention relates to a ferroelectric memory having a capacitor composed of a ferroelectric film.
- DRAM and flash memory-A ferroelectric memory having a ferroelectric capacitor in a memory cell has been developed as a semiconductor memory device having the advantages of EEPR0M.
- a ferroelectric memory operates by using a ferroelectric capacitor that uses a ferroelectric as an insulating material as a variable capacitance capacitor, and utilizes the fact that residual polarization remains even when the voltage applied to the ferroelectric capacitor is zero. In addition, data can be retained even when power is not supplied.
- a ferroelectric capacitor a ferroelectric material having a main composition of PZT (lead zirconate titanate) or a ferroelectric material having a bismuth layered belovskite structure such as SBT (bismuth strontium tantalate) is used. Available.
- the 1T1C cell is composed of one transfer transistor and one ferroelectric capacitor to hold one bit of information.
- 1T1C cells are used in ferroelectric memories for large-capacity applications because the memory cell size can be reduced.
- the 2T2C cell consists of two transfer transistors and two ferroelectric capacitors to hold one bit of information.
- the 2T2C cell stores complementary data in two ferroelectric capacitors, so the read margin can be increased. For this reason, 2T2C cells are used in ferroelectric memories for high reliability and reliability.
- One end of the ferroelectric capacitor of the above-described ferroelectric memory is connected to a bit line via a transfer transistor, and the other end is connected to a plate line.
- a plate line For example, in a read operation of a 2T2C ferroelectric memory, when a plate line is driven, the voltage difference between a pair of bit lines becomes an effective capacitance value of a pair of ferroelectric capacitors storing complementary data. It changes according to the difference. The voltage difference between the bit line pair is amplified by the sense amplifier, It is output to the outside as read data.
- This type of reading method is called a plate line driving method.
- the plate line is commonly connected to many ferroelectric capacitors, and has a large load capacitance. For this reason, in the read operation, the plate line drive method with a large CR delay has a disadvantage that the read access time is long.
- the present inventors have proposed a method of reading the capacitance difference of a ferroelectric capacitor as data without driving a plate line (capacity difference detection method) in order to shorten the read access time (non-patented). References (1)).
- FIG. 1 shows an outline of a capacitance difference detection read circuit using a 2T2C type ferroelectric memory cell.
- the memory cell MC has transfer transistors Ml and M2 composed of nMOS transistors and ferroelectric capacitors FC1 and FC2.
- the ferroelectric capacitor FC1 has one end connected to the bit line BL via the transfer transistor Ml and the other end connected to the plate line PL.
- One end of the ferroelectric capacitor FC2 is connected to the bit line XBL via the transfer transistor M2, and the other end is connected to the plate line PL.
- the gates of the transfer transistors Ml and M2 are connected to the word line WL.
- the arrows attached to the ferroelectric capacitors FC1 and FC2 indicate the polarization state. The upward arrow indicates a state where "logic 0" is stored. The downward arrow indicates a state where "logic 1" is stored.
- the sense amplifier SA has a pair of CMOS inverters whose inputs and outputs are connected to each other, and a pair of pMOS transistors P10 and P11.
- the source, drain and gate of the pMOS transistor P10 are connected to the power supply line VDD, the bit line BL, and the output CSC (hereinafter, also referred to as CSC signal) of the current source, respectively.
- the source, drain and gate of the pMOS transistor P11 are connected to the power supply line VDD, the bit line XBL, and the output CSC of the current source, respectively.
- the pMOS transistors P10 and P11 operate as current sources for supplying current to the bit line pairs BL and XBL, respectively.
- FIG. 2 shows a hysteresis loop of the pair of ferroelectric capacitors FC1 and FC2 shown in FIG.
- Plate line PL is driven from low level to high level, and both ends of ferroelectric capacitor
- the effective capacitance value CO becomes smaller because no polarization reversal is involved.
- the ferroelectric capacitor FC2 storing the logic 1 ⁇ involves polarization reversal, so that the effective capacitance C1 increases.
- FIG. 3 shows a read operation of the 2T2C type ferroelectric capacitor shown in FIG.
- the bit lines BL and XBL are precharged to the ground voltage VSS.
- the ferroelectric capacitors FC1 and FC2 are connected to the bit lines BL and XBL, respectively.
- the output CSC of the current source that is pulled up to the power supply voltage VDD is set to the DC bias value (low level). Due to the change of the CSC signal to the low level, the same amount of current flows through the bit lines BL and XBL via the pMOS transistors P10 and P11. At this time, the rising speed of the bit lines BL and XBL differs depending on the effective capacitance values of the ferroelectric capacitors FC1 and FC2. Specifically, the bit line BL connected to the ferroelectric capacitor FC1 having a small effective capacitance value is smaller than the bit line XBL connected to the ferroelectric capacitor FC2 having a large effective capacitance value. Get up early.
- the voltage on the bit lines BL and XBL continues to rise while the CSC signal is low. After a sufficient voltage difference occurs between the bit lines BL and XBL, the CSC signal changes again to the power supply voltage VDD, and the current source composed of the pMOS transistors P10 and P11 turns off. Thereafter, the sense amplifier power supplies SAP and SAN change to high level and low level, respectively, and the sense amplifier SA is activated.
- the voltage of the bit line BL rises to the voltage of the sense amplifier power supply SAP (for example, the power supply voltage VDD), and the voltage of the bit line XBL becomes the voltage of the sense amplifier power supply SAN (for example, the ground voltage VSS). Descend to During activation of the sense amplifier SA, the plate line PL is driven, and the original data is written back to the ferroelectric capacitors FC1 and FC2. Thereafter, the word line WL is deselected and the read operation is completed.
- the plate line PL is connected to the ferroelectric capacitor. Driven after data is read out from the FC1 and FC2. Therefore, data can be read from the memory cell MC to the bit lines BL and XBL without depending on the CR delay time of the plate line. Therefore, the data read time (the time from the read command to the output of the read data) can be reduced. Specifically, the data read time is about 40% shorter than the plate line drive method.
- FIG. 4 shows an outline of a read circuit of a capacitance difference detection method using a 1T1C type ferroelectric memory cell.
- the sense amplifier SA is the same as the sense amplifier SA shown in FIG.
- Each memory cell MC has a transfer transistor Ml composed of an nMOS transistor and a ferroelectric capacitor FC1.
- One end of the ferroelectric capacitor FC1 is connected to the bit line BLE or the bit line BL0 via the transfer transistor Ml, and the other end is connected to the plate line PL.
- the gates of the transfer transistors Ml of the memory cell MC are connected to different word lines WLE and WL0, respectively. That is, the memory cells MC connected to the complementary bit lines BLE and BL0, respectively, are not accessed at the same time.
- the reference memory cell RMC has a reference capacitor composed of the same four ferroelectric capacitors C0 and C1 as the ferroelectric capacitor FC1 of the memory cell MC, and two nMOS transistors N10 and Nil.
- the nMOS transistor N10 connects the reference capacitor to the bit line BLE when the reference line RWL0 is at a high level.
- the nMOS transistor Nil connects the reference capacitor to the bit line BL0 when the reference word line RWLE is high.
- the reference capacitor is configured by connecting in parallel two capacitance pairs in which a ferroelectric capacitor C0 storing “logic 0” and a ferroelectric capacitor C1 storing “logic 1 ⁇ ” are connected in series.
- the capacitance value of the reference capacitor is (C0 + C1) no 2.
- the reference capacitor is a ferroelectric capacitor FC1 that stores "logic 0” and a ferroelectric material that stores ⁇ logic 1 ⁇
- the capacitor has an intermediate capacitance value between the capacitance values of the capacitor FC1.
- the lead line WLE goes high and the When the memory cell MC connected to the cut line BLE is selected, the reference lead line R LE goes high to connect the reference capacitor to the bit line BL0. Similarly, if word line WL0 goes high and the memory cell MC connected to bit line BL0 is selected, reference word line RWL0 goes high to connect the reference capacitor to bit line BLE. . Then, as in FIG.
- the voltage of the bit line BLE (or BL0) that changes according to the capacitance of the ferroelectric capacitor FC1 and the bit line BL0 (or BL0) that changes according to the capacitance of the reference capacitor BLE) is amplified by the sense amplifier SA and output as read data.
- the capacitance difference detection method can shorten the read access time as compared with the plate line drive method. However, further reduction in read access time is required.
- a reference memory cell is configured using a plurality of ferroelectric capacitors.
- the relationship between the applied voltage and the capacitance value of the ferroelectric capacitor is non-linear.
- the ferroelectric capacitor FC1 storing ⁇ logic 0 ⁇ and the ferroelectric capacitor FC1 storing ⁇ logic 1 ⁇ shown in FIG. 2 have a capacitance with respect to a change in applied voltage. The value changes differently.
- the effective capacitance value of the reference memory cell will be smaller than (C0 + C1) 2.
- the read margin of the memory cell MC storing ⁇ logic 0 ⁇ is smaller than the read margin of the memory cell MC storing “logic 1”.
- the reference memory cell can be composed of a single ferroelectric capacitor that is larger than the ferroelectric capacitor FC1 of the memory cell.
- the effective capacitance value of the reference memory cell can be set between the capacitance values C0 and C1 shown in FIG.
- the ferroelectric memory when the number of readings (the number of rewritings) exceeds 10 to the 10th power, the ferroelectric material deteriorates, and the shape of the hysteresis loop shown in FIG. And the remanent polarization Q decreases.
- the remanent polarization value Q when a ferroelectric memory is mounted on a printed circuit board, the remanent polarization value Q temporarily decreases due to thermal fluctuations caused by the soldering process (heat treatment). The remanent polarization value Q is restored to the value before soldering by the first read operation after the heat treatment.
- a change in the remanent polarization Q before and after heat treatment lowers a read margin.
- the decrease in the remanent polarization Q is canceled out because the remanent polarization Q is reduced in both the memory cell MC and the reference memory cell. Therefore, the read margin hardly decreases.
- the remanent polarization value Q of the reference memory cell has recovered, so the read margin decreases and the data cannot be correctly read. Reading may not be possible.
- a ferroelectric memory (FIG. 4) in which a reference memory cell is shared by a plurality of memory cells MC connected to a plurality of read lines WLE and WL0, the number of accesses to the reference memory cell is limited to a maximum number of accesses to the memory cell MC. Double the number of times. For this reason, the material characteristics of the ferroelectric capacitor constituting the reference memory cell deteriorate faster than the material characteristics of the ferroelectric capacitor of the memory cell MC. As a result, the read margin decreases as the number of read operations increases, and the number of read operations (the number of rewrite operations) decreases.
- Patent Document (1) In order to prevent the read margin from decreasing in 1T1C ferroelectric memory, a technology has been proposed in which the average of the maximum and minimum voltages obtained in the read operation of the ferroelectric capacitor is set to the reference voltage ( Patent Document (1)).
- Patent Document (1) the number of memory cells MC connected to the word line WL is increased by one, and the increased memory cell MC is written with inverted data of data written to the memory cell MC connected to a predetermined bit line. It is.
- the operation of the sense pump cannot be started until the reference voltage is generated from the ferroelectric capacitor. Therefore, the access time becomes slow.
- An object of the present invention is to reliably read data from a memory cell even when the remanent polarization value of a ferroelectric capacitor is small and the voltage change of a bit line is small.
- the purpose is to maximize the data read margin for each logical value.
- Another object of the present invention is to secure a data reading margin and increase the number of times data can be rewritten even when a ferroelectric capacitor has deteriorated.
- Another object of the present invention is to reduce the read cycle time of a ferroelectric memory.
- externally supplied data is stored in the ferroelectric capacitor of the normal memory cell including the first memory cell.
- Inverted data of the first data stored in the first memory cell is stored in the ferroelectric capacitor of the second memory cell.
- the effective capacitance value of the ferroelectric capacitor differs depending on the logical value of the write data.
- the ferroelectric capacitor is charged.
- the voltage of the bit line connected to the ferroelectric capacitor having the small effective capacitance rises faster than the voltage of the bit line connected to the ferroelectric capacitor having the large effective capacitance. For this reason, the time until the voltage of the bit line exceeds the threshold voltage differs depending on the logical value of the data stored in the ferroelectric capacitor. In other words, the charging time of the ferroelectric capacitor differs depending on the logical value of the written data.
- the logical value of the data stored in the memory cell can be detected as a time difference.
- a circuit for detecting time can be formed more easily than a circuit for detecting voltage, and its accuracy is high. Therefore, it is necessary to determine the logical value of the data read from the normal memory cell to the bit line a predetermined time after the voltage of one of the bit lines connected to the first and second memory cells first exceeds the threshold voltage. As a result, data can be read reliably.
- the time difference can be reliably generated, so that even when the remanent polarization value of the ferroelectric capacitor is small, data can be reliably read from the memory cell. That is, the data read margin can be improved as compared with the case where the logical value of the data is detected based on the voltage difference. Further, even when the ferroelectric capacitor is deteriorated, a data read margin can be secured, so that the number of times data can be rewritten can be increased.
- the predetermined time is set such that the voltage of one of the bit lines connected to the first and second memory cells exceeds a threshold voltage and then the other of the bit lines connected to the first and second memory cells Is set to half of the period until the bit line voltage exceeds the threshold voltage.
- the predetermined time is set using, for example, a variable delay circuit.
- the period from when the voltage of one bit line exceeds the threshold voltage to when the predetermined time elapses is equal to the period from when the predetermined time elapses to when the voltage of the other bit line exceeds the threshold voltage.
- the sense amplifier differentially amplifies the difference between the voltage of each bit line and the threshold voltage and outputs the result as read data. That is, whether or not the bit line exceeds the threshold voltage is determined by the sense amplifier.
- the bit line voltage difference (the difference in logical value) is converted into a time difference between the transition edges of the read data output from the sense amplifiers.
- the latch signal is generated a predetermined time after the transition edge of the first read data (the center of both transition edges). That is, the latch signal is generated from the read data.
- the logic level of the read data is latched by the latch signal. Therefore, the data read margin can be maximized for each of the two logical values.
- An adjustment signal for reducing this difference is output. Therefore, when the output timing of the latch signal deviates from the center of both transition edges, the output timing can be corrected to the correct timing by adjusting the delay time of the variable delay circuit.
- the delay adjustment circuit has a counter that reverses the counting direction when the first period is longer than the second period and when the first period is shorter than the second period.
- the delay adjustment circuit outputs a counter value of the counter as an adjustment signal.
- the delay time of the variable delay circuit is adjusted by the adjustment signal. For this reason, the delay time of the variable delay circuit can be easily changed according to the increase or decrease of the counter value.
- a variable delay circuit includes a plurality of load capacitors connected to a latch signal generation path, and a switch connecting or disconnecting the load capacitance to the generation path according to a counter value. are doing. Therefore, the output timing of the latch signal can be easily changed using the counter.
- the capacitance value of the load capacitance is set to increase twice. Therefore, for example, the capacitance value can be increased or decreased by a predetermined value according to the weight of each bit of the binary counter. As a result, the delay time of the variable delay circuit can be adjusted at regular intervals.
- the latch signal is generated a predetermined time after the voltage of one of the bit lines exceeds a threshold voltage.
- the delayed latch signal is generated by further delaying the latch signal by the predetermined time.
- the latch signal and the delay latch signal are respectively generated by, for example, two serially connected variable delay stages whose delay times are always set to be equal.
- the first and second periods are equal.
- the first-arrival determination circuit of the variable delay circuit includes a read-out data output first from among the read-out data output from the sense amplifiers corresponding to the first and second memory cells. Select and output to the first variable delay stage. For this reason, the first variable delay stage can generate a latch signal based on the previously output read data without depending on the logical value of the data stored in the first and second memory cells.
- the late arrival determination circuit of the variable delay circuit outputs the read data output later from the read data output from the sense amplifiers corresponding to the first and second memory cells. Select and output to the delay adjustment circuit. For this reason, the delay adjustment circuit can optimally adjust the output timing of the latch signal based on the read data output later without depending on the logical value of the data stored in the first and second memory cells. .
- a write-back operation of writing data read from a memory cell to the memory cell again is started in response to a latch signal.
- the write-back operation is performed, for example, by driving a plate line connected to a ferroelectric capacitor. Since the write-back operation is started by the latch signal generated from the read data, the write-back operation can be started earlier and completed earlier. Therefore, the read cycle can be shortened.
- the predetermined time is set for each read operation.
- the set predetermined time is used in the next read operation. Therefore, even when the temperature fluctuates during the operation of the ferroelectric memory or the power supply voltage fluctuates, the output timing of the latch signal can always be set to be optimal.
- the bit lines are precharged to a predetermined voltage before a read operation.
- the voltage of the bit line can be increased by accurately reflecting the effective capacitance value of the strong dielectric capacitor of the first and second memory cells. Therefore, the output timing of the latch signal can be accurately set at the center between the output timing of the preceding read data and the output timing of the subsequent read data. As a result, the data read margin can be maximized for each of the two logical values.
- the voltage generation circuit generates a threshold voltage. Threshold voltage By generating the threshold voltage inside the ferroelectric memory, a desired threshold voltage can be easily generated.
- FIG. 1 is a circuit diagram showing an outline of a readout circuit using a capacitance difference detection method in a conventional 2T2C ferroelectric memory.
- FIG. 2 is a characteristic diagram showing a hysteresis loop of the ferroelectric capacitor shown in FIG.
- FIG. 3 is a timing chart showing a read operation of the 2T2C ferroelectric memory shown in FIG.
- FIG. 4 is a circuit diagram showing an outline of a read circuit using a capacitance difference detection method in a conventional 1T1C ferroelectric memory.
- FIG. 5 is a block diagram showing an embodiment of the ferroelectric memory of the present invention.
- FIG. 6 is a block diagram showing details of the current source control circuit CSC0N.
- FIG. 7 is a circuit diagram showing an example of the nonvolatile latch NVLT shown in FIG.
- FIG. 8 is a circuit diagram showing details of the threshold voltage generation circuit VGEN shown in FIG.
- FIG. 9 is a circuit diagram showing details of a main part of the memory core CORE shown in FIG.
- FIG. 10 is a block diagram showing details of the adaptive reference generation circuit ADLY shown in FIG.
- FIG. 11 is a block diagram showing details of the delay adjustment circuit DADJ shown in FIG.
- FIG. 12 is a timing chart showing a read operation of the ferroelectric memory of the present invention.
- Double circles in the figure indicate external terminals.
- the signal lines indicated by bold lines are composed of a plurality of lines.
- a part of the block to which the bold line is connected is composed of a plurality of circuits.
- FIG. 5 shows an embodiment of the ferroelectric memory of the present invention.
- Ferroelectric memory chips are formed on a silicon substrate using a CMOS process.
- This ferroelectric memory is, for example, a work memory of a mobile terminal such as a mobile phone.
- Used as The ferroelectric memory consists of a command buffer CMDB, command decoder CMDD, control circuit C0NT, address buffer ADB, row decoder RDEC, column decoder CDE word driver WD, precharge circuit PRE, current source control circuit CSC0N, and threshold voltage generation circuit. It has VGEN, memory core C0RE, plate driver PD, and data output buffer 0BF.
- Figure 1 mainly shows the circuits necessary for the read operation. Therefore, circuits such as a data input buffer and a write amplifier necessary for the write operation are omitted.
- Command buffer C receives command signals CMD such as chip select signal / CS, output enable signal / 0E and write enable signal / WE via the command terminal, and outputs the received signal to command decoder CMDD. .
- the command decoder CMDD decodes the operation mode indicated by the command signal, and outputs operation control signals (read control signal, write control signal, etc.) to the control circuit C0NT according to the decoded operation mode.
- the control circuit C0NT generates a timing signal for operating the plate driver PD, the word driver TO, the data output buffer 0BF, and the like according to the operation control signal.
- the address buffer ADB receives the address signal AD via the address terminal, and outputs the received signal to the row decoder RDEC and the column decoder CDEC.
- the decoder RDEC decodes the upper bits (row address) of the address signal to generate a row decode signal, and outputs the generated signal to the word driver TO.
- the column decoder CDEC generates a column decode signal by decoding the lower bits (column address) of the address signal, and outputs the generated signal to the column decoder column CDEC.
- the word driver WD selects a predetermined word line WL (LE, WL0) in response to a timing signal from the control circuit C0NT and a row decode signal from the row decoder RDEC.
- the selected word line WL changes from a low level to a high level.
- the plate driver PD selects a predetermined plate line PL in response to the timing signal from the control circuit C0NT and the row decode signal from the row decoder RDEC.
- the selected plate line PL changes from a low level to a high level for a predetermined period.
- the precharge circuit PRE precharges the bit lines BLE and BL0 to the ground voltage VSS before the read operation. Therefore, in the read operation, the bit lines BLE, BL0 Changes accurately according to the data stored in the memory cell MC.
- the current source control circuit CSC0N outputs a control signal CSC for turning on the current source CS of the memory core CORE according to a timing signal output from the control circuit C0NT during a read operation.
- the threshold voltage generation circuit VGEN generates a threshold voltage Vth used for a read operation.
- the threshold voltage Vth is a constant voltage (for example, 2.5 V) independent of the power supply voltage VDD.
- the power supply voltage VDD is, for example, 3.0 V as a standard, and is allowed from 2.7 V to 3.3 V as product specifications.
- the data output buffer 0BF selects 8 bits of the multi-bit read data read from the memory core CORE according to the column decode signal, and outputs the selected read data to the data input / output terminal I / O.
- Data input / output terminal I / O consists of 8 bits.
- the memory core CORE consists of a memory cell array ARY, a plurality of multiplexers MUX corresponding to bit line pairs consisting of bit lines BLE and BL0, a current source CS (current supply circuit), a sense amplifier SA, and a latch circuit LT. And an adaptive reference generation circuit ADLY.
- the sense amplifier SA, the latch circuit LT, and the adaptive reference generation circuit ADLY operate as a read control circuit.
- the memory cell array ARY has a plurality of memory cells MC arranged in a matrix, a plurality of lead lines WL (WLE, WL0) and a plurality of bit lines BLE and BL0 connected to the memory cells MC. ing.
- the memory cell MC is a 1T1C type memory cell and has a ferroelectric capacitor FC and a transfer transistor TR, as shown in FIG. 9 described later.
- the ferroelectric capacitor FC has one end connected to the bit line BLE (or BL0) via the transfer transistor TR, and the other end connected to the plate line PL.
- the gate of the transmission transistor TR is connected to the lead line WLE (or WL0).
- the suffix 'and ⁇ 0' of the word line WL and the bit lines BLE and BL0 indicate an even number and an odd number, respectively.
- the even number line WLE is connected to the even number line BLE via the memory cell MC.
- the odd word line WL0 is connected to the odd bit line BL0 via the memory cell MC, so that when one word line WL is selected, the bit line pair is always connected.
- One bit line (BLE or BL0) is connected to the memory cell MC. It is.
- the memory cell MC connected to the n sets of bit lines BLEl-BLEn and BLOl-BLOn operates as a normal memory cell that stores write data input from the data input / output terminal I / O.
- the memory cell MC connected to BL01-BLOn is also referred to as a first memory cell.
- the memory cell MC connected to the bit lines BLEn + l and BL0n + l is a second memory cell that stores the inverted data of the data (first data) stored in the first memory cell.
- the multiplexer MUX connects one of the bit lines BLE and BL0 to the sense amplifier SA according to the row decode signal.
- the current source CS applies a bias current to one of the bit lines BLE and BL0 selected by the multiplexer MUX during a predetermined period from the start of the read operation.
- the sense amplifier SA differentially amplifies the difference between the voltage of one of the bit lines BLE and BL0 selected by the multiplexer MUX and the threshold voltage Vth, and outputs the amplified voltage to the latch circuit LT.
- the latch circuit LT latches the read data output from the sense amplifier SA in synchronization with the complementary latch signals LAT and / LAT, and outputs the latched data to the data output buffer 0BF.
- the adaptive reference generation circuit ADLY consists of the data read from the first memory cell connected to the bit line BLEn (or BLOn) and the second memory cell connected to the bit line BLEn + 1 (or BL0n + l). Adjusts the generation timing of latch signals LAT and / LAT optimally according to the data read from As will be described later, the read margins of ⁇ logic 0 "and" logic 1 ⁇ are both maximized by the adaptive reference generation circuit ADLY.
- FIG. 6 shows details of the current source control circuit CSC0N shown in FIG.
- the current source control circuit CSC0N consists of pMOS transistors P21, P22, resistors 1R, 2R, 4R, nMOS transistors N21, N22, N23, and nonvolatile latches NVLT and AND gates corresponding to resistors 1R, 2R, 4R, respectively. And The numbers at the beginning of the resistors 1R, 2R, and 4R indicate the resistance ratio of these resistors. That is, the resistance values of the resistors 1R, 2R, and 4R are sequentially set to twice that of the other resistors.
- the pMOS transistor P21 forms a current mirror circuit with the pMOS transistor P31 forming the current source CS shown in FIG. 9 described later.
- the pMOS transistor P22 turns off while the control signal CNT is at a high level, and turns on while the control signal CNT is at a low level.
- the voltage of the control signal CSC is set to the power supply voltage VDD when the control signal CNT is at a low level, the read control signal RDP is at a low level, and the nMOS transistors N21, N22, and N23 are all off.
- the control signal CNT and the read control signal RDP are output from the control circuit C0NT, and change to a high level during a predetermined period during the read operation.
- the resistor 1R (or 2R, 4R) and the nMOS transistor N21 (or N21, N23) are connected in series between the output node of the control signal CSC and the ground line VSS.
- the gates of the nMOS transistors N21, N22 and N23 are connected to outputs CIN1, CIN2 and CIN3 of AND gates which receive the output of the nonvolatile latch NVLT and the read control signal RDP, respectively.
- the non-volatile latch NVLT always outputs a high level or a low level, respectively, depending on the latched logic value. However, it is prohibited that both nonvolatile latches NVLT output low level.
- a current to be supplied to the current mirror circuit is set according to the number of nMOS transistors N21, N22, N22 which are turned on during a predetermined period during the read operation. For example, when the control signals CIN1 and CIN2 are at a high level and the control signal CIN3 is at a low level, the nMOS transistors N21 and N22 are turned on and the nMOS transistor N23 is turned off. Therefore, a current corresponding to (VDD-VT) Z (R + 2R) flows through the current mirror circuit.
- VT is the threshold voltage (absolute value) of the pMOS transistor P21.
- the voltage of the control signal CSC is set to seven low levels in a predetermined period during the read operation in accordance with the 3-bit control signals CIN1, CIN2, and CIN3 output from the nonvolatile latch NVLT. Voltage. Therefore, the amount of current flowing through the current mirror circuit can be easily adjusted according to the data latched on the nonvolatile latch NVLT.
- the value stored in the nonvolatile latch NVLT can be set after manufacturing the ferroelectric memory. That is, the value of the current flowing through the current mirror circuit can be set after manufacturing the ferroelectric memory. Therefore, device characteristics can be changed according to the product specifications of the ferroelectric memory. Specifically, by setting a large current value, a ferroelectric memory having high-speed access can be obtained. By setting the current value small, a highly reliable ferroelectric memory with many rewrites can be obtained. One chip can be made into multiple products with different specifications according to the value stored in the nonvolatile latch NVLT, thereby reducing product costs.
- FIG. 7 shows an example of the nonvolatile latch circuit NVLT shown in FIG.
- the nonvolatile latch circuit NVLT includes a latch circuit LT2 having a switch for cutting off a power supply voltage VDD and a ground voltage VSS, two CMOS switches SW for writing data to the latch circuit LT2, and a plate line PL1, PL2.
- a pair of ferroelectric capacitors FC connected in series via the input node N of the latch circuit LT2 and a pair of ferroelectric capacitors FC connected in series via the output node NX of the latch circuit LT2 between the plate lines PL1 and PL2.
- the nMOS transistor N30 is controlled by the store signals ST01 and ST02, respectively.
- the nMOS transistor N31 is controlled by reset signals RES1 and RES2, respectively.
- the CMOS switch SW is controlled by sampling clock signals CK and CKX, respectively.
- the nonvolatile latch circuit NVLT is a known circuit, a detailed description is omitted.
- the latch circuit applied to the present invention is not limited to the nonvolatile latch circuit NVLT.
- the nonvolatile latch circuit may be constituted by a nonvolatile flip-flop circuit having a ferroelectric capacitor or a nonvolatile SRAM.
- the nonvolatile flip-flop circuit is configured by cascade-connecting two nonvolatile latch circuits NVLT or a nonvolatile latch circuit VNLT and a volatile latch circuit.
- Non-volatile SRAM has a memory cell with six transistors similar to the non-volatile latch circuit NVLT. It consists of four ferroelectric capacitors. By composing the latch circuit using a ferroelectric capacitor having the same structure as the ferroelectric capacitor of the memory cell MC, it is possible to prevent the manufacturing process from becoming complicated.
- FIG. 8 shows details of the threshold voltage generation circuit VGEN shown in FIG.
- the threshold voltage generation circuit VGEN has a bandgap reference VREF, an operational amplifier 0PAMP, and resistors Rl and R2, which are configured by a well-known CMOS circuit.
- the bandgap reference BGR generates a reference voltage (approximately 1.2 V) that is the voltage of the silicon band gap.
- the operational amplifier 0PAMP performs feedback control so that the divided voltage of the threshold voltage Vth by the resistors R1 and R2 is equal to the reference voltage, and generates a constant threshold voltage Vth (2.5 V).
- the reference voltage is always maintained at a constant value regardless of the operating temperature of the ferroelectric memory, changes in the power supply voltage, and changes in the process conditions in the ferroelectric memory manufacturing process. For this reason, the threshold voltage generation circuit VGEN can always generate a constant threshold voltage Vth.
- FIG. 9 shows details of a main part of the memory core CORE shown in FIG.
- the memory cell MC is an mC memory cell having a ferroelectric capacitor FC and a transfer transistor TR.
- One end of the ferroelectric capacitor FC is connected to the bit line BLE (or BL0) via the transfer transistor TR, and the other end is connected to the plate line PL.
- the gate of the transmission transistor TR is connected to the word line WLE (or WL0).
- the multiplexer MUX has an nMOS transistor and a pMOS transistor for each of the bit lines BLE and BL0.
- the nMOS transistor and the pMOS transistor are controlled by bit line selection signals BLSE and BLS0, respectively, which are a kind of input / output signal.
- the bit line select signal BLSE changes to high level when the even-numbered lead line WLE is selected.
- the bit line selection signal BLS0 changes to high level when the odd word line WL0 is selected.
- the current source CS has a pMOS transistor P31 that receives the control signal CSC output from the current source control circuit CSC0N by Gout.
- the current source CS applies a current corresponding to the low-level voltage to the bit line BLE (or BL0) while the control signal CSC is at a low level.
- the sense amplifier SA combines a differential amplification type amplifier and a common source type amplifier. It is configured.
- the differential amplifier is activated in response to the sense amplifier activation signal SAN, and differentially amplifies the voltage difference between the threshold voltage Vth and the bit line BLE (or BL0). Before the start of read operation, bit lines BLE and BL0 are precharged to ground voltage VSS.
- the threshold voltage Vth supplied to the sense amplifier SA is always generated.
- the voltages of the bit lines BLE and BL0 at the start of reading are lower than the threshold voltage Vth. Therefore, even if the sense amplifier SA is activated early in the read operation, no malfunction occurs. As a result, the activation timing of the sense amplifier SA can be set earlier, and the read access time can be reduced.
- the common-source amplifier converts the output voltage of the differential amplifier to the CMOS level and outputs it as a data signal DT (read data).
- the sense amplifier SA changes the data signal DT from the low level to the high level when the voltage of the bit line BLE (or BL0) exceeds the threshold voltage.
- the latch circuit LT has a latch composed of a pair of CMOS inverters, a CMOS switch that transmits a data signal DT to the latch, and an nMOS transistor that interrupts the latch feedback loop.
- the CMOS switch turns on when the latch signal LAT (/ LAT) is at a low level (high level) and turns off when the latch signal LAT (/ LAT) is at a high level (low level).
- the nMOS transistor conducts the loop when the latch signal LAT is high, and breaks the loop when the latch signal LAT is low.
- the latch circuit LT outputs the held data as a data signal D0UT. Note that a CMOS switch may be used instead of the nMOS transistor.
- the sense amplifiers SA corresponding to the bit lines BLEn and / BLEn output the data signals DTn and / DTn, respectively, as shown in FIG.
- the data signal DTn is output not only to the latch circuit LT but also to the adaptive reference generation circuit ADLY.
- the data signal / DTn is output only to the adaptive reference generation circuit ADLY.
- FIG. 10 shows details of the adaptive reference generation circuit ADLY shown in FIG.
- the adaptive reference generation circuit ADLY has a variable delay circuit VDLY and a delay adjustment circuit DADJ.
- the variable delay circuit VDLY has a pair of variable delay stages VDLY1 and VDLY2 connected in series, an AND gate, and an OR gate.
- the AND gate is used to synchronize the data signal DTn, / DTn, which changes slowly to a high level. As a result, a late arrival signal LATE which changes to a high level is output. That is, the AND gate operates as a late arrival determination circuit that selects a signal transmitted later from the data signals DTn and / DTn and outputs the selected signal to the delay adjustment circuit DADJ. The data signal transmitted later can be easily detected by the AND gate.
- the OR gate outputs a first-arrival signal EARY that changes to a high level in synchronization with a signal that changes to a high level early among the data signals DTn and / DTn. That is, the OR gate operates as a first-arrival determination circuit that selects the signal transmitted earlier from the data signals DTn and / DTn and outputs the selected signal to the variable delay stage VDLY1. The data signal transmitted earlier can be easily detected by the OR gate.
- the ferroelectric capacitor FC of the memory cell MC (first memory cell) connected to the bit line BLEn stores ⁇ logic 0 ⁇
- the ferroelectric capacitor FC of the memory cell MC (second memory cell) stores logic 1 ⁇ .
- the effective capacitance CO of the ferroelectric capacitor FC storing ⁇ logic 0 ⁇ is smaller than the effective capacitance C1 of the ferroelectric capacitor FC storing ⁇ logic 1 ⁇ .
- the sense amplifier SA changes its output to a high level when the voltage of the bit lines BLEn and / BLEn exceeds the threshold voltage Vth. Therefore, the rising edge timing of the first-arriving signal EARY output from the 0R gate is the time when the sense amplifier SA power V logic 0 "is detected.
- the rising edge timing of the late-arriving signal LATE output from the AND gate is equal to the sense timing. It is time to detect the SA logic V logic 1 ⁇ , that is, the rising edge timing of the early arrival signal EARY and the late arrival signal LATE means reading “logic ⁇ ” and ⁇ logic 1 ⁇ , respectively.
- the logic of the read data read to the bit lines BLE and BL0 is detected not as a voltage difference but as a time difference.
- the first variable delay stage VDLY1 delays the first-arrival signal EARY output from the OR gate by a predetermined time T1, and outputs the delayed signal as latch signals / LAT and LAT.
- the variable delay stage VDLY2 delays the latch signal LAT by a predetermined time T1, and outputs the delayed signal to the delay adjustment circuit DADJ as a delay latch signal LATD.
- the first and second variable delay stages VDLY1 and VDLY2 receive the same delay adjustment signal ADJ0-2 and are always set to the same delay time T1.
- Each of the variable delay stages VDLY1 and VDLY2 has three capacitors lCd, 2Cd and 4Cd connected to the transmission path of the first-arrival signal EARY (the transmission path of the latch signal LAT) via the nMOS transistors N41, N42 and N43. .
- the numbers at the beginning of the capacities lCd, 2Cd, and 4Cd indicate the capacity ratio of these capacities. That is, the capacitance values of the capacitances lCd, 2Cd, and 4Cd are sequentially set to twice the other capacitances.
- the gates of the nMOS transistors N41, N42, N43 receive the delay adjustment signals ADJ0-2 output from the delay adjustment circuit DADJ, respectively.
- the capacitance value added to the transmission path of the first incoming signal EARY is set to eight (0Cd_7Cd) according to the logical value of the 3-bit delay adjustment signal ADJ0-2.
- the delay adjustment circuit DADJ compares the arrival time of the delayed latch signal LATD obtained by delaying the first-arrival signal EARY with the variable delay stages VDLY1 and VDLY2 and the late-arrival signal LATE for each read operation. When the arrival of the delay latch signal LATD is earlier than the arrival of the late arrival signal LATE, the delay adjustment circuit DADJ changes the logical value indicated by the delay adjustment signal ADJ0-2 to ⁇ ⁇ 1 in order to increase the delay time of the variable delay circuit VDLY. "To increase.
- the delay adjustment signal ADJ0 corresponds to the lower bit
- ADJ2 corresponds to the upper bit
- the delay adjustment circuit DADJ changes the delay adjustment signals ADJ0-2 to increase the delay time of the variable delay circuit VDLY when the arrival of the delayed latch signal LATD is earlier than the arrival of the late arrival signal LATE.
- the delay adjustment signal ADJ0-2 is changed from binary "logic 01" to "logic 100"
- the capacitance value added to the transmission path of the first arrival signal EARY increases from "3CcT" to "4Cd”.
- the delay times of the delay stages VDLY1 and VDLY2 both increase by the same time.
- the delay adjustment circuit DADJ changes the delay adjustment signal ADJ0-2 to reduce the delay time of the variable delay circuit VDLY.
- the delay adjustment signal ADJ0-2 is changed from binary “logic 011” to "logic 010"
- the capacitance value added to the transmission path of the first-arrival signal EARY is reduced from "3Cd” to "2Cd” .
- Variable delay stages VDLY1 and VDLY2 have the same delay time T1 Decrease. Such adjustment of the delay time T1 is performed for each read operation until the arrival of the delay latch signal LATD coincides with the arrival of the late arrival signal LATE.
- the output timings of the latch signals LAT and / LAT become the rising edge timing of the first arrival signal EARY and the late arrival signal LATE. This is set at the center of the rising edge timing. That is, the output timings of the latch signals LAT and / LAT are set at the center between the time when the sense amplifier SA logic 0 is detected and the time when the logic 1 is detected. As a result, the bit for reading ⁇ logic 0 ⁇ is set.
- the read margin is maximized for ⁇ logic 0 ⁇ and ⁇ logic 1 ⁇ , respectively.
- the delay time is adjusted for each read operation. Therefore, even when the latch timing is shifted due to a change in the temperature during operation of the ferroelectric memory or a change in the power supply voltage VDD, the latch timing with the largest read margin can be restored. Further, even when the remanent polarization value changes due to deterioration of the ferroelectric capacitor FC, the read margin can always be maximized with respect to "logic 0" and "logic 1".
- FIG. 11 shows details of the delay adjustment circuit DADJ shown in FIG.
- the delay adjustment circuit DADJ determines the first arrival of the rising edge of the delay latch signal LATD and the late arrival signal LATE, and outputs the up signal UP or the down signal DOWN to the differential amplifier AMP and the up signal UP or the down signal DOWN. It has a counter COUNT that operates upon receiving.
- the differential amplifier AMP changes the up signal UP to a low level when the rising edge of the delay latch signal LATD is earlier than the rising edge of the late arrival signal LATE.
- the differential amplifier AMP changes the down signal DO west to a low level when the rising edge of the delay latch signal LATD is later than the rising edge of the late arrival signal LATE.
- the drains of the nMOS transistors receiving the up signal UP and the down signal DOWN, respectively, are connected to each other via a key part transistor that slightly reduces the detection sensitivity of the differential amplifier AMP.
- the counter COUNT counts up in response to the falling edge of the up signal UP, counts down in response to the falling edge of the down signal DOWN, and outputs the counter value as the delay adjustment signal ADJ0-2. It is an up-down counter. That is, the counter COUNT counts down when the first period P1 is longer than the second period P2, and counts up when the first period P1 is shorter than the second period P2.
- both the up signal UP and the down signal DOWN are at a high level, the read operation is not executed and the differential amplifier AMP is in a non-operation state. At this time, the counter COUNT does not operate.
- both the up signal UP and the down signal DO are low, the rising edge timings of the delay latch signal LATD and the late arrival signal LATE are equal. Specifically, when the difference between the rising edges of the delay latch signal LATD and the late arrival signal LATE is smaller than the quantization error (minimum delay adjustment time) of the variable delay circuit VDLY, the up signal UP and the down signal D0 ⁇ , Are set to low level. At this time, the counter COUNT does not operate.
- FIG. 12 shows a read operation of the ferroelectric memory of the first embodiment.
- Data is externally written to the memory cell MC in advance.
- the memory cell MC (second memory cell) connected to the bit lines BLEn and BLEn contains inverted data of the data stored in the memory cell MC (first memory cell) connected to the bit lines / BLEn and / BLOn. Has been written.
- the control circuit C0NT shown in FIG. 5 When the address signal AD for selecting the memory cell MC from which the read command and data are to be read is supplied to the ferroelectric memory, the control circuit C0NT shown in FIG. 5 outputs the control signal CNT and the read control signal RDP ( Figure 12 (a, b)).
- the mode driver WD changes the word line WLE (or WL0) corresponding to the address signal AD to a high level.
- the row decoder RDEC changes the bit line selection signal BLSE (or BLS0) to a high level in accordance with the address signal AD (FIG. 12 (c)).
- the current source control circuit CSC0N shown in FIG. 6 changes the control signal CSC to a low level in synchronization with the rising edge of the read control signal RDP (FIG. 12 (d)).
- control The low level voltage of the signal CSC is set according to the data stored in the nonvolatile latch NVLT.
- the change of the control signal CSC turns on the current source CS, and the current is supplied to the bit line BLE (or BL0).
- the voltage of the bit line BLE (or BL0) gradually increases according to the remanent polarization value (capacitance value) of the ferroelectric capacitor FC connected by the word line WLE (or WL0) (Fig. 12 (e)). .
- the voltage of the bit line connected to the ferroelectric capacitor FC storing ⁇ logic 0 ⁇ is higher than the voltage of the bit line connected to the ferroelectric capacitor FC storing “logic 1 ⁇ . Ascend fast.
- the sense amplifier SA shown in FIG. 9 changes the data signal DT to a high level when the voltage of the bit line connected to the ferroelectric capacitor FC storing ⁇ logic 0 ⁇ exceeds the threshold voltage Vth ( Figure 12 (f)).
- another sense amplifier SA changes the data signal DT to a high level when the voltage of the bit line connected to the ferroelectric capacitor FC storing "logic 1" exceeds the threshold voltage Vth (see FIG. 1 2 (g)).
- the OR gate of the adaptive reference generation circuit ADLY shown in Fig. 10 outputs the first-arrival signal EARY in synchronization with the first-arrival data signal DT (either DTn or / DTn) (Fig. 12 (h) ).
- the AND gate synchronizes with the late-arriving data signal DT (DTn or / DTn) and outputs the late-arriving signal LATE (Fig. 12 (i)).
- the adaptive reference generation circuit ADLY delays the first-arrival signal EARY by the two variable delay stages VDLY1 and VDLY2, and outputs it as a delay latch signal EARYD (FIG. 12 (j)).
- FIG. 12 shows an example in which the delay time of the variable delay circuit VDLY is optimally set. Therefore, the rising edge of the delayed latch signal LATD is at the same timing as the rising edge of the late arrival signal LATE.
- the period P1 from the rising edge of the first arrived data signal DT to the output of the latch signals LAT, / LAT is the period P2 from the output of the latch signals LAT, / LAT to the rising edge of the second arrived data signal DT. be equivalent to. Further, these periods Pl and P2 are also equal to the delay time T1 of the variable delay stages VDLY1 and VDLY2.
- the variable delay stage VDLY1 outputs latch signals LAT and / LAT (FIG. 12 (k)).
- the output timings of the latch signals LAT and / LAT are timings that maximize the read margin with respect to "logic 0" and "logic 1".
- the latch circuit LT shown in FIG. The read data latched in synchronization with the latch signals LAT and / LAT is output as the data signal D0UT (Fig. 12 (1)).
- the delay adjustment circuit DADJ increases the logic value of the adjustment signal ADJ0-2 from “011 ⁇ ” to "100", thereby increasing the delay time of the variable delay circuit VDLY (FIG. 1). 2 (m)).
- the output timing of the latch signals LAT and / LAT is slightly delayed in the next read operation.
- the control circuit C0NT changes the control signal CNT to a low level after the sense amplifier SA outputs the data signal DT (FIG. 12 (n)).
- the control signal CSC changes to the high level due to the low-level control signal CNT (Fig. 12 (o)).
- the sense amplifier activation signal SAN changes to low level, and the sense amplifier SA is deactivated (FIG. 12 (p)).
- the plate dryer PD shown in FIG. 5 changes the plate line PL to a high level for a predetermined period according to the address signal AD (FIG. 12 (q)).
- the voltage of the bit line BLE (or BL0) changes according to the remanent polarization value of the ferroelectric capacitor FC (Fig. 12 (r)).
- the bit line connected to the ferroelectric capacitor FC storing ⁇ logic 1 ⁇ changes to low level, and the bit line connected to the ferroelectric capacitor FC storing ⁇ logic 0 ⁇ Change to a high level.
- the data read from the memory cell MC is written back to the read memory cell MC.
- the word line WLE (or WL0) changes to low level, and the read operation is completed (Fig. 12 (s)).
- the rising timing of the plate line PL is set, for example, a predetermined time after the rising edge of the latch signal LAT.
- the latch signal LAT is generated from the data signal DT by the adaptive reference generation circuit ADLY.
- the plate line PL can always start up at an optimum time from the output of the data signal DT regardless of the operating environment.
- write back can be started while the read data D0UT is being output, and the read cycle time can be reduced.
- the rise timing of the plate line PL is set by a delay circuit or the like formed in the control circuit C0NT. For this reason, it is necessary to set the startup timing of the plate line PL in consideration of the worst operating voltage and operating temperature, and write-back cannot be started during output of the read data D0UT.
- the logical value of the data stored in the memory Detected as a difference.
- a circuit for detecting time can be formed more easily than a circuit for detecting voltage, and its accuracy is high. Therefore, data can be reliably read out with a simple circuit.
- the output timing of the latch signals LAT and / LAT is set at the center of the timing when ⁇ Logic 0 ⁇ is output and the timing when ⁇ Logic 1 ⁇ is output.
- the data read margin can be maximized for each of the two logical values.
- the read margin can always be maximized.
- the delay adjustment for each read operation even when the temperature fluctuates during the operation of the ferroelectric memory or the power supply voltage VDD fluctuates, the output timing of the latch signals LAT and / LAT is always optimized. Can be set.
- the capacitance values of the capacitances lCd, 2Cd, and 4Cd that constitute the variable delay stages VDLY1 and VDLY2 are set to be twice as large, and the capacitances lCd, 2Cd, and 4Cd connected to the transmission path of the first-arrival signal EARY are converted to binary. Set according to the counter value of the counter. Therefore, the delay time can be adjusted at equal intervals according to the weight of the binary counter.
- the delay time of the variable delay stages VDLY1 and VDLY2 is adjusted according to the timing difference between the transition edge of the delayed latch signal LATD in which the first-arrival signal EARY is delayed by the variable delay stages VDLY1 and VDLY2 and the transition edge of the late-arrival signal LATE. Then, the latch signals LAT and / LAT were output from the connection node of the two variable delay stages VDLY1 and VDLY2 having the same delay time T1. Therefore, the time difference between the first period P1 and the second period P2 is easily and reliably eliminated, and the transition edge of the latch signal LAT / LAT is changed between the transition edge of the first arrival signal EARY and the transition edge of the second arrival signal LATE. Can be set in the center.
- the write-back operation is started.
- the back operation can be performed repeatedly during data reading. As a result, the readout total time can be reduced.
- the OR gate is formed as the first-arrival determination circuit, the data signal transmitted earlier can be easily detected. Further, since an AND gate is formed as the late arrival determination circuit, a data signal transmitted later can be easily detected.
- the precharge circuit PRE that precharges the bit lines BLE and BL0 before the read operation is formed, the voltage of the bit lines BLE and BL0 is raised by accurately reflecting the effective capacitance value of the ferroelectric capacitor FC. be able to.
- the threshold voltage generation circuit VGEN is formed in the ferroelectric memory, a desired threshold voltage Vth without fluctuation can be easily generated.
- the present invention is applied to a ferroelectric memory chip.
- the present invention is not limited to such an embodiment.
- the present invention may be applied to a ferroelectric memory core embedded in a system LSI.
- the variable delay circuit VDLY may delay a complementary signal using a differential amplification type delay circuit.
- the output timings of the latch signals LAT and / LAT can be made equal, and the latch circuit LT can operate at high speed.
- the data can be reliably read by detecting the logical value of the data stored in the memory cell as a time difference. Specifically, the logic value of data read from the normal memory cell to the bit line a predetermined time after a voltage of one of the bit lines connected to the first and second memory cells first exceeds the threshold voltage, By judging, data can be reliably read. Even if the voltage change of the bit line is small, the time difference can be reliably generated, so that even when the residual polarization value of the ferroelectric capacitor is small, data can be reliably read from the memory cell. That is, the read margin of the electric data can be improved. Furthermore, the ferroelectric capacity Even when the deterioration has occurred, the data read margin can be secured, so that the number of times data can be rewritten can be increased.
- the data read margin can be maximized for each of the two logical values by setting the timing for determining the logical value of the data at the center of the read timing of the two logical values. That is, the read margin can be improved.
- the output timing of the latch signal deviates from the center of both transition edges, the output timing can be corrected to the correct timing by adjusting the delay time of the variable delay circuit.
- the delay time of the variable delay circuit can be easily changed according to the increase or decrease of the counter value.
- the output timing of the latch signal can be easily changed using a counter. Since the capacitance value can be increased or decreased for each predetermined value according to the weighting of each bit of the counter, the delay time of the variable delay circuit can be adjusted at regular intervals.
- the first and second periods can be easily and reliably equalized, and the latch signal can be set at the center of both transition edges. .
- the first variable delay stage is provided by the first-arrival determination circuit of the variable delay circuit, and the first variable delay stage does not depend on the logical values of the data stored in the first and second memory cells, and the read data output first is output. Can generate a latch signal.
- the late adjustment circuit determines the delay adjustment circuit based on the read data output later without depending on the logical value of the data stored in the first and second memory cells.
- the output timing of the latch signal can be adjusted optimally.
- the write-back operation since the write-back operation is started by the latch signal generated from the read data, the write-back operation can be started earlier and completed earlier. Therefore, the read cycle can be shortened.
- the output timing of the latch signal is always optimized even when the temperature fluctuates or the power supply voltage fluctuates during the operation of the ferroelectric memory. Can be set.
- the voltage of the bit line is reduced by the effective voltage of the ferroelectric capacitors of the first and second memory cells.
- the capacitance value can be accurately reflected and increased. Therefore, the output timing of the latch signal can be accurately set at the center between the output timing of the preceding read data and the output timing of the subsequent read data.
- a threshold voltage having a desired value can be easily generated by generating the threshold voltage inside the ferroelectric memory.
Landscapes
- Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- Dram (AREA)
Abstract
Description
Claims
Priority Applications (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/JP2003/006601 WO2004107350A1 (ja) | 2003-05-27 | 2003-05-27 | 強誘電体メモリ |
| AU2003241803A AU2003241803A1 (en) | 2003-05-27 | 2003-05-27 | Ferroelectric memory |
| JP2005500197A JP4157553B2 (ja) | 2003-05-27 | 2003-05-27 | 強誘電体メモリ |
| US11/188,104 US7266009B2 (en) | 2003-05-27 | 2005-07-25 | Ferroelectric memory |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/JP2003/006601 WO2004107350A1 (ja) | 2003-05-27 | 2003-05-27 | 強誘電体メモリ |
Related Child Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| US11/188,104 Continuation US7266009B2 (en) | 2003-05-27 | 2005-07-25 | Ferroelectric memory |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2004107350A1 true WO2004107350A1 (ja) | 2004-12-09 |
Family
ID=33485767
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/JP2003/006601 Ceased WO2004107350A1 (ja) | 2003-05-27 | 2003-05-27 | 強誘電体メモリ |
Country Status (4)
| Country | Link |
|---|---|
| US (1) | US7266009B2 (ja) |
| JP (1) | JP4157553B2 (ja) |
| AU (1) | AU2003241803A1 (ja) |
| WO (1) | WO2004107350A1 (ja) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9093168B2 (en) | 2012-03-05 | 2015-07-28 | Fujitsu Semiconductor Limited | Nonvolatile latch circuit and memory device |
Families Citing this family (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4177131B2 (ja) * | 2003-02-06 | 2008-11-05 | ローム株式会社 | 論理演算回路、論理演算装置および論理演算方法 |
| US20050190597A1 (en) * | 2004-02-27 | 2005-09-01 | Yoshihisa Kato | Semiconductor device |
| US7733729B2 (en) * | 2004-04-01 | 2010-06-08 | Nxp B.V. | Thermally stable reference voltage generator for MRAM |
| JP2006344289A (ja) * | 2005-06-08 | 2006-12-21 | Toshiba Corp | 強誘電体記憶装置 |
| JP4801977B2 (ja) * | 2005-11-18 | 2011-10-26 | 株式会社東芝 | 半導体記憶装置 |
| TWI266338B (en) * | 2005-12-01 | 2006-11-11 | Via Tech Inc | Output circuit of SRAM |
| US7633786B2 (en) | 2006-04-18 | 2009-12-15 | Micron Technology, Inc. | Couplings within memory devices and methods |
| US7903477B2 (en) * | 2008-02-29 | 2011-03-08 | Mosaid Technologies Incorporated | Pre-charge voltage generation and power saving modes |
| US7933138B2 (en) * | 2009-01-30 | 2011-04-26 | Texas Instruments Incorporated | F-RAM device with current mirror sense amp |
| JP2013190893A (ja) * | 2012-03-13 | 2013-09-26 | Rohm Co Ltd | マルチタスク処理装置 |
| US9336873B2 (en) * | 2013-12-02 | 2016-05-10 | Intel Corporation | Apparatus for time domain offset cancellation to improve sensing margin resistive memories |
| CN106105037B (zh) * | 2014-01-21 | 2019-03-29 | 拉迪安特技术公司 | 利用铁电电容器的非易失性计数器 |
| EP3507804A4 (en) | 2016-08-31 | 2020-07-15 | Micron Technology, INC. | FERROELECTRIC MEMORY CELLS |
| EP3507805B1 (en) | 2016-08-31 | 2025-10-01 | Micron Technology, Inc. | Apparatuses and methods including ferroelectric memory and for operating ferroelectric memory |
| CN109087674A (zh) | 2017-06-14 | 2018-12-25 | 萨摩亚商费洛储存科技股份有限公司 | 铁电内存及其数据读取、写入与制造方法和电容结构 |
| US10867675B2 (en) | 2017-07-13 | 2020-12-15 | Micron Technology, Inc. | Apparatuses and methods for memory including ferroelectric memory cells and dielectric memory cells |
| US10504576B2 (en) * | 2017-12-19 | 2019-12-10 | Micron Technology, Inc. | Current separation for memory sensing |
| JP6796681B2 (ja) * | 2019-05-13 | 2020-12-09 | ウィンボンド エレクトロニクス コーポレーション | 半導体記憶装置 |
| US11996137B2 (en) * | 2021-05-21 | 2024-05-28 | Taiwan Semiconductor Manufacturing Company, Ltd. | Compute in memory (CIM) memory array |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH11185465A (ja) * | 1997-12-17 | 1999-07-09 | Rohm Co Ltd | 強誘電体メモリ |
| JP2002032984A (ja) * | 2000-07-17 | 2002-01-31 | Matsushita Electric Ind Co Ltd | 強誘電体記憶装置 |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1325500B1 (en) * | 2000-09-25 | 2005-12-28 | Symetrix Corporation | Ferroelectric memory and method of operating same |
| JP3866913B2 (ja) | 2000-11-21 | 2007-01-10 | 富士通株式会社 | 半導体装置 |
| US6914839B2 (en) * | 2001-12-24 | 2005-07-05 | Intel Corporation | Self-timed sneak current cancellation |
-
2003
- 2003-05-27 JP JP2005500197A patent/JP4157553B2/ja not_active Expired - Fee Related
- 2003-05-27 AU AU2003241803A patent/AU2003241803A1/en not_active Abandoned
- 2003-05-27 WO PCT/JP2003/006601 patent/WO2004107350A1/ja not_active Ceased
-
2005
- 2005-07-25 US US11/188,104 patent/US7266009B2/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH11185465A (ja) * | 1997-12-17 | 1999-07-09 | Rohm Co Ltd | 強誘電体メモリ |
| JP2002032984A (ja) * | 2000-07-17 | 2002-01-31 | Matsushita Electric Ind Co Ltd | 強誘電体記憶装置 |
Non-Patent Citations (1)
| Title |
|---|
| YADOLLAH ESLAMI ET AL.: "A differential-capacitance read syndrome for FeRAMs", 2002 SYMPOSIUM ON VLSI CIRCUITS DIGEST OF TECHNICAL PAPERS, 13 June 2002 (2002-06-13) - 15 June 2002 (2002-06-15), pages 298 - 301, XP002971858 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9093168B2 (en) | 2012-03-05 | 2015-07-28 | Fujitsu Semiconductor Limited | Nonvolatile latch circuit and memory device |
Also Published As
| Publication number | Publication date |
|---|---|
| AU2003241803A1 (en) | 2005-01-21 |
| US20060083049A1 (en) | 2006-04-20 |
| US7266009B2 (en) | 2007-09-04 |
| JPWO2004107350A1 (ja) | 2006-07-20 |
| JP4157553B2 (ja) | 2008-10-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP4157553B2 (ja) | 強誘電体メモリ | |
| KR100276569B1 (ko) | 강유전메모리장치 | |
| KR101071212B1 (ko) | 반도체 메모리 | |
| US8213253B2 (en) | Semiconductor memory | |
| JP2004134026A (ja) | 半導体記憶装置及びその制御方法 | |
| US7130211B2 (en) | Interleave control device using nonvolatile ferroelectric memory | |
| JP2017511950A (ja) | Framメモリにおけるインプリント低減のための回路及び方法 | |
| WO2012019861A1 (en) | Morphing memory architecture | |
| KR100506458B1 (ko) | 멀티비트 제어 기능을 갖는 불휘발성 강유전체 메모리 장치 | |
| US20240379151A1 (en) | Bit line sense amplifier and memory device including the same | |
| JP2006107560A (ja) | 半導体メモリ装置 | |
| KR100527539B1 (ko) | 고속 센싱을 위한 불휘발성 강유전체 메모리 장치 | |
| US7397685B2 (en) | Semiconductor memory device having error checking and correcting circuit | |
| US7751222B2 (en) | Semiconductor memory device | |
| JP2008171525A (ja) | 半導体記憶装置 | |
| KR100765872B1 (ko) | 강유전체 메모리 | |
| US7729157B2 (en) | Semiconductor storage device | |
| US7808805B2 (en) | Column address control circuit capable of selectively enabling sense amplifier in response to column addresses | |
| US6967860B2 (en) | Ferroelectric memory device and control method thereof | |
| JP4477629B2 (ja) | 強誘電体メモリ | |
| US6995998B2 (en) | Nonvolatile ferroelectric memory device having a multi-bit control function | |
| JP2009271991A (ja) | 半導体記憶装置 | |
| US7304882B2 (en) | Circuits for driving FRAM | |
| JP2024006885A (ja) | 半導体記憶装置および半導体記憶装置の読み出し方法 | |
| JP2025129759A (ja) | 半導体記憶装置および半導体記憶装置の制御方法 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| AK | Designated states |
Kind code of ref document: A1 Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NI NO NZ OM PH PL PT RO RU SC SD SE SG SK SL TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW |
|
| AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): GH GM KE LS MW MZ SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IT LU MC NL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG |
|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
| WWE | Wipo information: entry into national phase |
Ref document number: 2005500197 Country of ref document: JP |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 11188104 Country of ref document: US |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 1020057014223 Country of ref document: KR |
|
| WWP | Wipo information: published in national office |
Ref document number: 1020057014223 Country of ref document: KR |
|
| WWP | Wipo information: published in national office |
Ref document number: 11188104 Country of ref document: US |
|
| 122 | Ep: pct application non-entry in european phase |