JPS5947395B2 - Bias charge injection method in charge-coupled devices - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a bias charge injection method for stably injecting a constant amount of bias charge into a charge-coupled device.

A charge-coupled device is composed of a semiconductor substrate having an insulating film on the surface and a plurality of drive electrodes provided on the insulating film, and is used to discharge charges accumulated along the interface between the semiconductor substrate and the insulating film. , has a function of transmitting data using drive pulses applied to the drive electrodes.

Such charge-coupled devices function as shift registers and have a wide range of applications in the fields of digital devices and analog devices. Such a charge-coupled device is also called a CCD, and its principle is described in, for example, W in B-S-T-J, April 1970, Vol. 49, No. 4. It is described in detail in the paper "Charge Coupled Semiconductor Devices" by S. Paul and Y. E. Smith. By the way, when a charge-coupled device is used as an analog device, it transfers a charge proportional to the signal it handles, and when it is used as a digital device, the "l" of the signal or information it handles corresponds to, for example, "with charge". The information ``0'' corresponds to ``no charge'' and is transferred.

In either case, charge-coupled devices have the property that as the number of transfers increases, there is a charge attenuation called transfer loss, and the signal attenuates accordingly.

One effective means for reducing this charge attenuation is a method called fat zero. In this method, a fixed amount of bias charge is injected and transferred to a charge-coupled device regardless of the presence or absence of signal charge. It is widely known that the presence of this bias charge reduces the attenuation of signal charges. However, in the conventional bias charge injection method, even if there is a slight variation in the bias voltage for injecting bias charges, the bias charge fluctuates greatly, making it difficult to inject a fixed amount of bias charge. The drawback was that it was difficult.

Figure 1 shows a conventional method of injecting bias charges into a charge-coupled device.
2 is an insulating film provided on the semiconductor substrate, and 3, 4, 5, . . . are drive electrodes provided on the insulating film. The drive electrodes 3, 4, 5, . The drive pulse generators 6, 7, and 8 each have a phase difference of 120°. Further, the input gate electrode 9 and the output gate electrode 10 are provided via the insulating film 2 so as to partially overlap with the diffusion layers 11 and 12 in which n-type impurities are diffused, respectively. Here, the diffusion layer 11 is directly grounded, and the diffusion layer 12 is grounded via a load resistor 13 and a power source 14. Since three-phase positive pulses are constantly applied to the drive electrode groups 3, 4, 5, . . . by the drive pulse generators 6, 7, 8, the period during which the positive pulses are applied to the electrodes 3 If a positive pulse is applied from the input signal generator 16 to the input gate electrode 9, electrons (signal charge ) is injected. The signal charge is sequentially transferred to the right by the above-mentioned 6, 7, and 8 and passes under the output gate electrode 10,
It is guided to the reverse biased diffusion layer 12, discharged through the load resistor 13, and detected outside the substrate as a potential difference across the load resistor 13.

By the way, in charge-coupled devices, as mentioned above,
Since charge attenuation occurs due to transfer loss, the signal charge decreases as the number of charge transfers increases.

The main cause of this is thought to be a trap or the like that captures signal charges existing at the interface between the semiconductor substrate 1 and the insulating film 2. Therefore, if this trap is filled in advance with bias charges, the signal charges are transferred without being captured by the traps, thereby reducing transfer loss and improving transfer efficiency. Next, the conventional method for injecting the bias charge will be described. A method of injecting a bias charge sufficiently smaller than the signal charge by adding a DC bias charge to the input gate electrode 9 regardless of the presence or absence of an input signal. has been carried out.

In Fig. 2, an input signal voltage IN is applied to the input gate electrode,
Actual and experimental results regarding signal charge injection characteristics when signal charge Q is injected are shown. In the transition region of Fig. 2, IN
The relationship between Q and Q is acute. This means that if you apply a bias voltage of 0.7 to the input gate voltage and try to inject about 10% of the signal charge as the bias charge,
This means that a bias voltage variation of only ±0.1 V causes the bias charge amount to vary greatly from 0% to 55% of the signal charge amount. Therefore, the conventional bias charge injection method has the disadvantage that it is difficult to stably inject a constant amount of bias charge. Another problem arises when the conventional bias charge injection method is applied to a charge regeneration circuit.

In other words, as described above, after the signal charge is transferred many times, the signal charge amount attenuates, so when the signal charge attenuates to a certain amount of charge, a charge regeneration circuit is required to regenerate the input signal charge amount. . The method of injecting bias charges is also a technique required in the charge regeneration section of the charge regeneration circuit. FIG. 3 shows a conventional reproducing circuit, in which a plurality of drive electrodes 22,
23, 24, 25, 26, and 27 are arranged, and among the electrode groups, 22 and 25, 23 and 26, and 24 and 27 are connected. Moreover, the electrodes 22, 23, 24 each have 12
Drive pulse generators 28, 29, 3 with a phase difference of 0°
It is grounded via 0. The output gate electrode 31 is provided via the insulating film 21 so as to partially overlap the diffusion layers 32 and 33, and is grounded via the drive pulse generator 34. Here, the diffusion layer 33 is grounded via a power source 45 with a potential difference D, and the diffusion layer 32 is connected to the input gate electrode 3 via a wiring 36.
7 is connected. Further, the control gate electrode 38 is provided via the insulating film 21 so as to partially overlap the grounded diffusion layer 39, and is grounded via the drive pulse generator 40. Note that the electrodes 22, 23, and 24 connect the terminal end of the charge-coupled device to which the signal charge has been transferred to the electrodes 25, 26, and 2.
7 indicates the starting end of the next-stage charge-coupled device to which the newly reproduced signal charge is transferred. The operation of such a charge regeneration circuit is summarized as follows.
When signal charges are transferred to the signal charge detection diffusion layer, a potential change occurs depending on the amount of transferred signal charges. This potential change is applied to the input gate electrode of the next-stage charge-coupled device, and signal charges are regenerated and injected from another independent charge injection diffusion layer through under the input gate electrode. Normally, when a potential change does not occur, that is, no signal charge is transferred, the signal charge is injected into the next stage charge-coupled device, and when a potential change occurs, that is, when the signal charge is transferred, No signal charge is injected into the next stage charge-coupled device. That is, information is inverted via the charge regeneration circuit. Next, the operation of the conventional reproducing circuit will be explained in more detail with reference to FIG. 1, that is, in a state where no positive pulse is applied to the drive electrodes 24 and 25, a positive pulse is applied to the output gate electrode 31 from the drive pulse generator 34, and the diffusion layer 32 is precharged (precharged) to a positive voltage VD. ).

The precharge voltage VD is applied via the wiring 36. It is applied to the cover electrode 37. During this applied period,
A positive pulse is applied to the electrode 24 from the drive pulse generator 30. If the signal charge is transferred below the electrode 24, the signal charge is guided to the diffusion layer 32 at the fall of the positive pulse.
Here, if the sum of the capacitances of the diffusion layer 32, wiring 36, and input gate electrode 37 is C, and the amount of signal charge transferred to the diffusion layer 32 is Qt, then the diffusion layer 32 has a potential change of Qt/C. arise. Therefore, the voltage of the diffusion layer 32 is VD-Qt/
It drops to C. The voltage VD-Qt/C is the input gate 3
If the threshold voltage is lower than the threshold voltage of 7, even if a positive pulse is applied from the drive pulse generator 40 to the control gate electrode 38 during the period when the positive pulse is applied from the drive pulse generator 28 to the electrode 25, the diffusion layer No signal charge is injected from 39.
If the signal charge is not transferred to the diffusion layer 32, no potential change occurs in the diffusion layer 32 due to the signal charge, and the diffusion layer 32
And input gate electrode 37 holds precharge voltage VD. Since the precharge voltage VD is set higher than the threshold voltage of the input gate 37, if a positive pulse is applied to the control gate electrode 38 during the period in which the positive pulse is applied to the electrode 25, the electrode will be removed from the diffusion layer 39. The signal charges are injected under the electrode 25 through the electrodes 38 and 37, and the signal charges are reproduced. The regenerated signal charge is sent to the drive pulse generator 28.
, 29, and 30 are sequentially transferred to the right. Therefore, in the charge regeneration circuit shown in FIG. 3, the signal charge (information) is inverted and transmitted to the next stage charge-coupled device. In order to apply a bias voltage to the input gate electrode 37 and inject a certain amount of bias charge in the charge regeneration circuit shown in FIG. The precharge voltage VD must be set to a value slightly higher than the threshold voltage of the input gate 37, depending on the amount of bias charge desired to be injected. However, the amount of signal charge changes depending on fluctuations in the peak values of the positive pulses generated by the drive pulse generators 28, 29, and 30, or the input signal series. Therefore, in order to stably inject a constant amount of bias charge, the precharge voltage V must be adjusted according to the change in the amount of signal charge.
D must be controlled, which is very difficult to achieve. Therefore, conventionally, as shown in FIG. 4, a new input gate electrode 4L power source 42 is provided, and the electrode 2 passes from the diffusion layer 39 under the electrode 41 regardless of the presence or absence of signal charges.
A method has been adopted in which a fixed amount of bias charge is injected into the lower part of the cell.

However, this method also has the same operating principle as the bias charge injection method explained in Fig. 1, and the bias voltage (
Since the amount of bias charge varies greatly due to fluctuations in the power supply 42), there are disadvantages in that not only a constant amount of bias charge cannot be stably injected, but also the area occupied by the charge regeneration circuit increases. The present invention has been made in view of the above points, and in order to eliminate the drawbacks of the conventional bias charge injection method described above, the input gate electrode is used as a storage electrode when bias charge is injected, and the signal charge is When injecting, by using it as a normal gate in a time-sharing manner, it is possible to minimize bias charge fluctuations in response to bias voltage fluctuations and to stably supply a constant amount of bias charge. It is an object of the present invention to provide a good bias charge injection method.

A detailed explanation will be given below with reference to the drawings. FIGS. 5A and 5B are diagrams for explaining one embodiment of the present invention, and FIG. Figure B is a diagram for explaining the pulse timing of the structure in this case.

The difference between the conventional FIG. 1 and the present invention FIG.
This is because a control gate electrode 51 is newly provided which is grounded through the control gate electrode 51. FIG. 5B shows the driving pulse generators 6, 7,
voltage waveforms 6', 7', and 8' applied to the drive electrodes 3, 4, and 5 from the input signal generator 16, the voltage waveform 16' applied to the input gate electrode 9 from the input signal generator 16, and the voltage waveform 16' applied to the input gate electrode 9 from the drive pulse generator 50 5 is a diagram showing a voltage waveform 50'' applied to the control gate electrode 51. Next, the bias charge injection method according to the present invention will be explained using FIG. 5. That is, in FIG. Pulse generator 6
The control gate electrode 51 is in a state where no positive pulse is applied.
When a positive pulse is applied from the drive pulse generator 50 to the input gate electrode 9 and a positive pulse (bias voltage) is also applied to the input gate electrode 9, the voltage at the input gate electrode 9 is determined by the voltage applied to the input gate electrode 9 and the capacitance of the input gate electrode 9. Bias charges are injected from the diffusion layer 11 through below the electrode 51. Next, the voltage applied to the control gate electrode 51 is set to 0V, and after applying a positive pulse to the drive electrode 3,
When the voltage applied to the input gate electrode 9 is O, the bias charge stored under the input gate electrode 9 is transferred to the electrode 3.
It is transferred downward and becomes a bias charge. In this case, the input gate electrode 9 has a function as a storage electrode. Further, the amount of bias charge is usually selected to be about 1/10 of the amount of signal charge. Next, with the bias charge stored under the electrode 3, a positive pulse (input signal voltage) is again applied to the control gate electrode 51 at the timing indicated by "1" in FIG. 5B, and within the application period, When a positive pulse is applied to the input gate electrode 9,
Signal charges pass below the electrodes 51 and 9 from the diffusion layer 11 to the electrode 3.
It is injected downward, joins with the already stored bias charge, and is sequentially transferred to the right by the three-phase positive pulses from 6, 7, and 8. In this case, the input gate electrode 9 functions as a normal gate. Even if no positive pulse is applied to the input gate electrode 9 or no positive pulse is applied to the electrode 51 at the timing indicated by "0" in FIG. 5B, the bias charge stored under the electrode 3 is held and sequentially transferred to the right by the three-phase positive pulses from 6, 7, and 8. As described above, when the input gate electrode 9 is injected with bias charges, it is used as a storage electrode.
When injecting signal charges, a predetermined amount of bias charges can be stably supplied by using the gate as a normal gate in a time-division manner. It is known that if the bias charge amount is about 5 to 20% of the signal charge amount, the signal charge transfer efficiency of the charge-coupled device is improved and the S/N ratio of the signal charge is improved. To achieve this using the present invention, the following method is used. The method is to make the area of the input gate electrode 9 a fraction of the area of the drive electrodes 3, 4, and 5, or to make the bias voltage a fraction of the drive pulse voltage, or both. This method is to be used. As explained above, according to the present invention, when injecting bias charges, the input gate electrode 9 is used as a storage electrode, so the amount of bias charges stored under the input gate electrode 9 corresponds to the amount of saturation charge shown in FIG. do. Therefore, even with variations in the peak value of the positive pulses applied to the input electrode 9 and the control gate electrode 51 or manufacturing variations in the threshold voltage of the control gate 51, the amount of bias charge stored under the input gate electrode 9 will be It is not easily affected and its fluctuations are small. Therefore, according to the method described above, a predetermined bias charge can be stably injected, and a more stable bias charge injection method than the conventional method can be easily realized. The present invention is also useful when applied to a charge regeneration circuit, and an example of the structure when the present invention is applied to such a charge regeneration circuit as another embodiment of the present invention is shown in FIG. 6A. .

The difference between the conventional charge regeneration circuit shown in FIG. 6A and FIG. 3 is that a drive pulse generator 60 is newly provided and connected to the electrode 25. FIG. 6B shows the drive pulse generator 28,
Drive electrodes 22, 23 and 2 from 29, 30 and 60 respectively
Voltage waveform 28゛, 2 applied to 6, 24 and 27, 25
9″, 30″, 60″, a voltage waveform 34″ applied from the drive pulse generator 34 to the output gate electrode 31, and a voltage waveform 40″ applied from the drive pulse generator 40 to the control gate electrode 38. 6 is a diagram. The timing in the shaded part in the diagram is a new timing added according to the present invention, and is different from the conventional timing. Next, using FIG. 6, the charge regeneration circuit according to the present invention is The operation when applied is explained below.

That is, in FIG. 6, a positive pulse is applied from the drive pulse generator 34 to the output gate electrode 31 without applying a positive pulse to the drive electrodes 24 and 25, and the diffusion layer 32 is precharged to a positive voltage D (precharge). I'll keep it. The precharge voltage D is applied to the input gate electrode 3 via the wiring 36.
7. During the application period, the control gate electrode 38
When a positive pulse is applied from the drive pulse generator 40 to the input gate electrode 37, the precharge voltage D and the electrode 3
A bias charge determined by a capacitance of 7 is injected from the diffusion layer 39 through below the electrode 38. The precharge voltage VD
is set to be higher than the threshold voltage of the input gate 37 when signal charges are not transferred to the diffusion layer 32, and lower than the threshold voltage of the input gate 37 when signal charges are transferred to the diffusion layer 32. be done. Next, the control gate electrode,
When the voltage applied to the input gate electrode 38 is set to 0V and a positive pulse is applied to the electrode 25 from the drive pulse generator 60, the input gate electrode 3
All the charges stored under the electrode 25 are transferred to the bottom of the electrode 25 and become bias charges. The subsequent operation is similar to that of a conventional reproducing circuit. That is, if the signal charge l is not transferred to the diffusion layer 32, the voltage of the diffusion layer 32, that is, the voltage applied to the input gate electrode 37, will maintain the precharge voltage VD, so when a positive pulse is applied to the control gate electrode 38, the signal charge will be diffused. It is injected from the layer 39, passing under the electrodes 38 and 37 and under the electrode 25, and together with the bias charge that has already been stored, is sequentially transferred to the right by the aforementioned 28, 29, and 30. When the signal charge is transferred to the diffusion layer 32, the voltage of the diffusion layer 32, that is, the voltage applied to the input gate electrode 37 becomes lower than the threshold voltage of the input gate 37, so a positive pulse is applied to the electrode 38. Even if the voltage is applied, no new signal charges are injected from the diffusion layer 39 through below the electrodes 38, 37, and only the bias charges already stored under the electrodes 25 are sequentially moved to the right by the above-mentioned 28, 29, 30. be transferred. As described above, when injecting bias charges/charges, the input gate electrode 37 is used as a storage electrode, and when signal charges are injected, it is used as a normal gate in time division to stably supply a predetermined amount of bias charge. I can do it.
The amount of accumulated charge (bias charge) when the input gate electrode 37 is used as a storage electrode is equal to the precharge voltage VD.
Since it is determined by the capacitance of the electrode 37, it is not affected in any way by fluctuations in signal charges transferred to the diffusion layer 32. Furthermore, when bias charge is injected using the input gate electrode 37 as a storage electrode, the amount of bias charge stored under the input gate electrode 37 is the saturation charge amount shown in FIG. 2, as described in the explanatory text of FIG. corresponds to Therefore, the input gate electrode 3
The variation in the amount of bias charge stored under 7 is significantly smaller than in the conventional method. Furthermore, the amount of bias charge is hardly affected by variations in the threshold voltage of the control gate 38, and a predetermined amount of bias charge can be stably supplied. In the explanation of FIG. 6, a new drive pulse generator 60 is provided for convenience; however, in reality, the drive pulse generator 28
, 29, 30 to the drive electrodes 22, 23, 24 are voltage waveforms 28', 29', 30' that overlap each other, so the electrode 25 is connected to the grounded drive pulse 28. and make it work. Therefore, when the present invention is applied to the charge regeneration circuit shown in FIG. 3, only the change 40' shown in FIG. 6B is required, and there is no need to change the structure of the charge regeneration circuit at all. Therefore, as explained above, according to the method of the present invention, in a charge-coupled device in which a signal charge is injected by applying a human signal voltage to the input gate electrode, a bias voltage that is difficult to achieve with the conventional bias charge injection method can be applied. It has become possible to stably inject charge. In other words, the present invention uses the input gate electrode as a storage electrode when injecting bias charges and as a normal gate when injecting signal charges, and uses it in a time-division manner to control fluctuations in the input signal voltage, precharge voltage, and control gate voltage. It has become possible to realize a bias charge injection method that is significantly less affected by manufacturing variations in gate threshold voltage or fluctuations in signal charge amount. Further, the present invention can be realized by relatively simple structure and timing changes. Furthermore, according to the present invention, bias charges can be stably supplied, so when applied to a shift register, even if there are fluctuations in the input signal voltage, control gate voltage, and manufacturing variations in the threshold voltage of the control gate, there will be no signal/noise. It can operate stably with little effect on ratio, and furthermore, when applied to serial memory, the signal /
The memory operation can be performed stably with little influence on the noise ratio. It is also extremely useful when increasing the number of bits in a shift register, increasing the number of bits between charge regeneration circuits in serial memory, or trying to obtain an output signal with good reproducibility from an analog input signal. have advantages. In the embodiments of the present invention, charge-coupled devices using n-channel three-phase drive with a surface channel structure have been described, but the invention is not limited to these, and may also apply to charge-coupled devices with l-phase, two-phase, or four-phase drive, or P-channel drive. It is clear from the above description that it can be widely applied.

[Brief explanation of drawings]

Figure 1 is a structural diagram of a conventional charge-coupled device, Figure 2 is the input characteristics of a charge-coupled device, Figure 3 is a structural diagram of a conventional charge regeneration circuit, and Figure 4 is the bias in a conventional charge regeneration circuit. FIG. 5 is a structural diagram showing a charge injection method, and FIG. 5 is a structural diagram and waveform diagram when applied to the input section of FIG. 1 as an embodiment of the present invention. FIG. FIG. 2 is a structural diagram and a waveform diagram when the present invention is applied to a charge regeneration circuit of FIG. 1, 20... P-type semiconductor substrate, 2, 21...
... Insulating film, 3, 4, 5 ... Drive electrode, 6,
7, 8... Drive pulse generator, 9...
Input gate electrode, 10... Output gate electrode, 2
2, 23, 24, 25, 26, 27... drive electrode, 28, 29, 30... drive pulse generator,
31... Output gate electrode, 32... N-type diffusion layer for taking in signal charges, 36...
A wiring connecting the diffusion layer 32 and the input gate electrode 37, 37.
...Input gate electrode, 38...Control gate electrode, 41...Newly provided input gate electrode, 51...Newly provided control gate electrode.

Claims (1)

[Claims] 1. A semiconductor substrate having an insulating film on its surface, a plurality of drive electrodes provided on the insulating film, drive pulse generation means for supplying drive pulses to these drive electrodes, and an input gate electrode for applying a signal voltage. A charge-coupled device characterized in that the input gate electrode is time-divisionally used as a storage electrode when injecting bias charges and as a normal gate when injecting signal charges. bias charge injection method in