JP5132262B2 - Back-illuminated linear image sensor, driving method thereof, and manufacturing method thereof - Google Patents

Description

  The present invention relates to a back-illuminated linear image sensor used in the field of remote sensing, for example, a driving method of the linear image sensor, and a manufacturing method.

  A large number of image sensors have been developed in which a large number of photodetectors are arranged in an array on a semiconductor substrate, and a signal charge readout circuit and an output amplifier are provided on the same substrate. In remote sensing, a linear image sensor with photodetectors arranged in a one-dimensional array is mounted on an artificial satellite, etc., and a two-dimensional image of the ground surface is obtained by making the direction perpendicular to the array coincide with the traveling direction of the satellite. Take a picture. In order to improve the image resolution, it is desirable to make the pixel pitch in the photodetector as small as possible, but doing so reduces the area of the photodetector. Therefore, there is a problem that the amount of light incident on the photodetector is reduced and the S / N ratio is deteriorated.

  As a clever means for improving the S / N ratio, a TDI (Time Delay and Integration) image sensor has been developed. The TDI system uses an FFT (full frame transfer) CCD (Charge Coupled Devices), which is a two-dimensional image sensor, and improves the S / N ratio by synchronizing the charge transfer timing with the movement timing of the subject image. This is a readout method of a CCD image sensor. In the case of remote sensing, TDI operation can be realized by adjusting the charge transfer in the vertical direction to the moving speed of the satellite. That is, when M stages of TDI operations are performed with a vertical CCD, the accumulation time is effectively M times. Therefore, the sensitivity is improved M times, and the S / N ratio is improved to √M times.

  In many cases, an image sensor for visible light performs imaging by making light incident from the surface side of the chip of the sensor. Incident light is photoelectrically converted inside the Si substrate to generate signal charges. In the above-described FFT type CCD and FT (frame transfer) type CCD, light is incident through a polysilicon electrode that controls vertical charge transfer. Therefore, there is a problem that light in the short wavelength region is absorbed by the polysilicon electrode, and the sensitivity of the CCD is lowered. As a countermeasure, a back-illuminated image sensor in which light is incident from the back side of the sensor chip has been proposed.

  For example, Patent Document 1 discloses a conventional X-ray image sensor in which a photocathode is disposed on the back side of a substrate. This sensor makes X-rays incident from the front surface side of the substrate, but is configured as a back-illuminated FT-CCD for visible light region light generated on the photocathode. In the back-illuminated CCD, it is necessary to efficiently collect signal charges generated on the light incident surface (substrate back surface) side of the Si substrate with a CCD provided on the front surface side of the Si substrate. Therefore, the Si substrate thickness of the photoelectric conversion portion is thinly formed to about several tens of μm. At this time, if charges are mixed into a readout circuit, such as a horizontal CCD or an output amplifier, crosstalk occurs, which needs to be prevented. In the example of Patent Document 1, only the photoelectric conversion portion is selectively thinned using the anisotropy of Si etching, and the readout circuit portion emits visible light so as to leave the original Si substrate thickness (about 300 μm). It is shut off.

  The invention of Patent Document 2 forms a shielding member for the beveled portion (the slope portion of the recess on the backside of the substrate) when the photoelectric conversion portion is thinned by Si anisotropic etching in the back-illuminated image sensor. By doing so, light incidence to the bevel portion is prevented.

JP-A-11-204072 JP 2002-231913 A

  However, the conventional back-illuminated image sensor has a problem that the boundary between the light incident area and the light-shielding area on the back surface of the substrate cannot be formed with high accuracy. For example, as in Patent Document 1, when it is intended to provide a light shielding effect by the thickness of the Si substrate after Si anisotropic etching, the film thickness of the bevel portion changes continuously. Therefore, the boundary portion between the light incident region and the light shielding region has a band shape with a certain width and the transmittance gradually changes, and the boundary line becomes blurred.

  Further, the bevel portion formed by anisotropic etching is the (111) plane of Si, and its opening edge (bottom side) is formed in parallel to the (100) direction of the substrate. On the other hand, the direction of the photodetector array formed on the substrate surface side does not necessarily coincide with the (100) direction of the substrate.

  In general, in a wafer process, a circuit pattern is aligned with a surface orientation of a substrate with reference to an orientation flat or notch (hereinafter referred to as an orientation flat) provided on a Si wafer. The orientation flat or the like is formed in the wafer manufacturing process, but the manufacturing tolerance of the standard wafer orientation flat or the like is about ± 1 deg. For this reason, an angle shift of about 1 deg at the maximum occurs between the edge of the opening on the back side of the substrate and the detector array on the front side of the substrate.

  In remote sensing, a long chip having a pixel array length of several tens of mm or more is often used. Therefore, for example, when the pixel array length is 60 mm and the angle shift is 1 deg, a shift of 1 mm or more occurs at both ends of the pixel array in the distance from the opening edge to the pixel. As a result, there is a problem that the uniformity of sensitivity between pixels is lowered.

  Further, since the bevel portion is formed obliquely reflecting the (111) plane of Si, a ghost due to reflected light is generated when light enters here. The invention of Patent Document 2 prevents light from entering the bevel portion. For this purpose, a method of attaching a shielding member having an opening window to the chip is adopted. However, even if an attempt is made to form the boundary between the light incident region and the light shielding region by such a method, it is difficult to align the boundary line between the light incident region and the light shielding region with the edge of the photoelectric conversion region with high accuracy. . Therefore, there is a problem that light scattered at the edge of the opening of the shielding member becomes stray light and enters the light incident region.

  Furthermore, a method of depositing and patterning a metal film or the like on the incident surface on the back surface of the substrate and using this as a light shielding film is also conceivable. However, there is also a problem that manufacturing using this method is difficult in terms of preventing cracking of the thinned portion and patterning with a high step structure.

  As a result, when the TDI image sensor is a back-illuminated type, there is a problem in that noise charges generated by light incident on a region other than the photoelectric conversion unit are mixed into the signal charges and crosstalk occurs. In order to eliminate the noise charge, it is necessary to devise such as using a mechanical shutter, and continuous imaging such as moving image imaging is difficult with a configuration using a mechanical shutter together.

  The present invention has been made to solve the above-described problems, and a back-illuminated linear image sensor capable of reducing crosstalk generated by light incident on a region other than a photoelectric conversion unit, the linear An object of the present invention is to provide an image sensor driving method and a manufacturing method thereof.

In order to achieve the above object, the present invention is configured as follows.
That is, the back-illuminated linear image sensor according to the first aspect of the present invention includes the charge transfer element according to light incident from a light incident region formed on the back side of a semiconductor substrate in which a plurality of charge transfer elements are arranged in a line on the surface. A linear image sensor that performs back-illuminated TDI processing for transmitting signal charges generated in the above-described configuration, and includes a light receiving unit, a horizontal transfer unit, a high-speed transfer unit, and a drive unit. The light receiving unit includes a first charge transfer elements arranged in a first direction along the surface of the semiconductor substrate opposite to the light incident region, transfers the generated the signal charge to said first direction To do. The horizontal transfer unit includes a second charge transfer element arranged in the second direction along the surface so as to face a light shielding region other than the light incident region on the back surface side, Forward in the direction. The high-speed transfer unit is a third charge arranged between the light receiving unit and the horizontal transfer unit along the surface and arranged in the first direction at a boundary between the light incident region and the light shielding region. It has a transfer device, wherein said third charge transfer device, the light of the signal charge supplied incident from the transfer vital light incident region in the first direction the third charge transfer from the light receiving portion It enters the element and generates a noise charge. The driving unit causes the high-speed transfer unit to perform charge transfer without performing charge transfer to the light receiving unit, discharges only the noise charge, and horizontally transfers the signal charge transferred from the light receiving unit. High-speed transfer operation control to be transferred to the unit.

  According to the back-illuminated linear image sensor of the first aspect of the present invention, the high-speed transfer unit is provided between the light receiving unit and the horizontal transfer unit, and the drive unit that controls the drive is further provided. Therefore, the drive unit first causes the high-speed transfer unit to discharge noise charges generated in the high-speed transfer unit, and then transfers the signal charge generated in the light-receiving unit to the horizontal transfer unit. Therefore, crosstalk generated by light incident on a region other than the light receiving portion can be reduced, and moving image imaging can be performed without a mechanical shutter. The high-speed transfer unit is located at the boundary between the light incident region and the light shielding region formed on the back surface of the semiconductor substrate, and generates noise charges. However, as described above, since noise charges are first discharged, it is not necessary that the boundary portion be formed in parallel with the array of charge transfer elements constituting the light receiving unit. Therefore, it is possible to make the back-illuminated linear image sensor easier to manufacture than in the past.

A back-illuminated linear image sensor according to an embodiment of the present invention, a driving method of the linear image sensor, and a manufacturing method will be described below with reference to the drawings. In each figure, the same or similar components are denoted by the same reference numerals.
In the back-illuminated linear image sensor described below, a CCD (charge coupled device) is used as a charge transfer device, but the present invention is not limited to this, and a CMOS can also be used. Further, in order to improve the S / N ratio in the linear image sensor, the back-illuminated linear image sensor of the present embodiment also adopts the above-mentioned TDI (Time Delay and Integration) method, and FFT (full frame transfer). A type CCD or an FT (frame transfer) type CCD is used.

Embodiment 1 FIG.
FIG. 1 is an element plan view showing a circuit configuration of a back-illuminated TDI-CCD as an example of a back-illuminated image sensor 101 according to Embodiment 1 of the present invention. FIG. 2 schematically shows a cross section of the back-illuminated TDI-CCD along the line AA ′ shown in FIG.
The back-illuminated linear image sensor 101 includes, as a basic configuration, a light receiving unit 50, a high-speed transfer unit 60, a horizontal transfer unit 70, and a drive unit 80, and includes the light receiving unit 50, the high-speed transfer unit 60, and the horizontal transfer unit 70. Signal charges generated by light incident from the light incident region 19 provided on the back surface of the formed semiconductor substrate 30, that is, the back surface 12 a of the bulk Si substrate 12 are sent out. The drive unit 80 is a component that is provided separately from or on the semiconductor substrate that forms the light receiving unit 50, the high-speed transfer unit 60, the horizontal transfer unit 70, and the like. The drive unit 80 includes a first drive unit 81 that drives the CCD that constitutes the high-speed transfer unit 60, a second drive unit 82 that drives the CCD that constitutes the light receiving unit 50, a horizontal transfer unit 70, and an accumulation described later. A third drive unit 83 that drives the unit 6, the accumulation control gate 7, and the reset gate 8.

  On the front side of the semiconductor substrate 30 facing the back surface, a light receiving unit 50 in which TDI pixels 1 formed by the FFT or FT type CCD are arranged in a two-dimensional array is provided. In the present embodiment, in the present embodiment, the TDI pixels 1 are formed in 5 rows and 8 columns, and in each column of the light receiving portion 50, the TDI pixels 1 are in a vertical direction 91 (downward in the drawing) corresponding to the first direction. Arranged along. In the present embodiment, the high-speed transfer unit 60 includes vertical CCDs 3 formed in 3 rows and 8 columns, and the vertical CCDs 3 are arranged along the first direction 91 in each column. A storage unit 6, a storage control gate 7, and a reset gate 8 are formed corresponding to each column of the high-speed transfer unit 60. In the horizontal transfer unit 70, a horizontal direction (right side of the drawing) 92 corresponding to the second direction is formed. Horizontal CCDs 5 are arranged in 1 row and 8 columns along the line.

In the back-illuminated linear image sensor 101 configured as described above, the signal charge generated by the light receiving unit 50 by the above-described incident light and subjected to time delay integration constitutes the horizontal transfer unit 70 via the high-speed transfer unit 60. Is transferred to the horizontal CCD 5 in the vertical direction 91. The signal charges transferred to the horizontal transfer unit 70 are transferred in the horizontal direction 92 by the horizontal CCD 5 and read from the output amplifier 9.
In FIG. 1, the TDI pixels 1 of the light receiving unit 50 are illustrated in 5 rows and 8 columns. However, this is only an example, and can be arranged in a plurality of rows, one or a plurality of columns.

  As described above, between the light receiving unit 50 and the horizontal transfer unit 70, the high-speed transfer unit 60, the storage unit 6, the storage control gate 7, and the reset gate 8 including the vertical CCD 3 group are provided. As shown in FIG. 2, the frame 10 and the frame 11 shown by dotted lines in FIG. 1 are in a concave light incident region 19 formed by thinning a part of the Si substrate 12 from the back surface 12 a side of the Si substrate 12. It shows an edge. Further, the storage unit 6, the storage control gate 7 and the reset gate 8 are referred to as a storage control unit. Further, as will be described later, according to the back-illuminated linear image sensor 101 of the present embodiment, the high-speed transfer unit 60 needs to be accurately aligned with the end of the light receiving unit 50 at the boundary between the light incident region 19 and the light shielding region. Will disappear. In order to clarify this point, FIG. 1 intentionally illustrates the frame 10 and the frame 11 in a state where the frame 10 and the frame 11 are inclined with respect to the array of the light receiving unit 50 and the like. In FIG. 1, the CCD is shown in three lines for the high-speed transfer unit 60, but the present invention is not limited to this, and there may be a plurality of lines, and the number of lines is related to the area of the boundary part. Determined by

The back-illuminated image sensor 101 will be described with reference to FIG.
On the surface side of the SOI substrate 15 composed of the bulk Si substrate 12, the BOX oxide film 13, and the Si epitaxial layer 14, a readout circuit including a transfer channel 16 and a gate electrode 17 is formed. Here, 17a is a TDI transfer gate, 17b is a vertical CCD transfer gate, 17c is a storage gate, 17d is a storage control gate, and 17e is a horizontal CCD transfer gate included in the gate electrode 17. Also included in the transfer channel 16 is a channel portion of the TDI pixel 1, 16 b is a channel portion of the vertical CCD 3, 16 c is a channel portion of the storage portion, and 16 e is a channel portion of the horizontal CCD 5. The light receiving section 50 is composed of the TDI transfer gate 17a and the channel portion 16a of the TDI pixel 1, and the high-speed transfer section 60 is composed of the vertical CCD transfer gate 17b and the channel section 16b of the vertical CCD 3. The storage gate 17c, the storage control gate The storage unit 6 and the storage control gate 7 are configured by 17d and the channel portion 16c of the storage unit, and the horizontal transfer unit 70 is configured by the horizontal CCD transfer gate 17e and the channel portion 16e of the horizontal CCD 5.

  Reference numeral 18 denotes a protective film made of an oxide film or the like. Further, as described above, the light incident region 19 is formed by Si anisotropic etching from the substrate back surface 12a side. Here, the bevel portion 20 corresponding to the boundary portion at the edge of the light incident region 19 is a slope reflecting the Si (111) surface. Reference numeral 21 denotes an etching mask made of an oxide film or a nitride film when the light incident region 19 is patterned.

  With the above-described configuration, as shown in FIGS. 1 and 2, the light receiving unit 50 is disposed corresponding to the light incident region 19, and the high-speed transfer unit 60 is disposed corresponding to the bevel portion 20, so that the light incident is performed. The storage unit 6, the storage control gate 7, the reset gate 8, and the horizontal transfer unit 70 are arranged corresponding to the light shielding region 19c corresponding to the back surface 12a other than the region 19.

The operation of the back-illuminated image sensor 101 having such a configuration will be described below.
In the back-illuminated image sensor 101, light is incident from the back surface 12a side during imaging. In the light incident region 19, the bulk Si 12 is removed to reduce the thickness of the epitaxial layer 14, and the epitaxial layer 14 has a thickness of about 10 to 20 μm. Therefore, in the light receiving unit 50, signal charges are generated according to incident light, and the generated signal charges are accumulated in the channel 16a of the TDI pixel 1. On the other hand, the storage unit 6 and the horizontal CCD 5 are shielded from light by bulk Si 12 having a film thickness of several hundreds of μm and do not generate photocharges. In the high-speed transfer unit 60 located between them, the Si film thickness at the portion corresponding to the high-speed transfer unit 60 continuously changes from 10 μm to several hundred μm, so that the transmittance for visible light also changes continuously. Therefore, photocharges are also generated in the high-speed transfer unit 60, and the photocharges generated by the light incident on the high-speed transfer unit 60 are accumulated in the transfer channel 16b.

When the photocharge integrated in the transfer channel 16b is mixed with the signal charge generated in the light receiving section 50, crosstalk occurs, and a driving method for preventing this is necessary.
FIG. 3 is a potential diagram of the back-illuminated TDI-CCD 101 according to the first embodiment of the present invention, and shows the potential pattern of the transfer channel 16 in time series. Hereinafter, a method of driving the back-illuminated TDI-CCD 101 according to the first embodiment of the present invention will be described with reference to FIGS. In summary, the driving method performs one-stage TDI transfer in the light receiving unit 50 every imaging cycle, and the high-speed transfer unit 60 transfers the signal charge 22 transferred from the light receiving unit 50 to the horizontal CCD every imaging cycle. The high-speed transfer is performed, and the noise charge 23 generated in the high-speed transfer unit 60 is discharged immediately before the high-speed transfer of the signal charge 22. Such a driving method is controlled by the driving unit 80 and executed for the light receiving unit 50, the high-speed transfer unit 60, the storage unit 6, the storage control gate 7, the reset gate 8, and the horizontal transfer unit 70. The

  In the TDI pixel 1 of the light receiving unit 50, the signal charges 22 due to the incident light are integrated, and the signal charge amount is in the order of 22a, 22b,..., 22e in proportion to the number of TDI-CCDs constituting the light receiving unit 50. Become more. Further, since the high-speed transfer section 60 is located corresponding to the bevel portion 20 corresponding to the boundary between the light incident area 19 and the light shielding area 19c, the high-speed transfer section 60 also generates and accumulates photocharges. However, the photocharge generated in the high-speed transfer unit 60 is noise charge 23 (noise charges 23a to 23c) other than the signal charge 22. As shown in FIG. 3A, after one imaging cycle is completed, the signal charges 22 and the noise charges 23 integrated in the light receiving unit 50 and the high-speed transfer unit 60 are respectively transferred in the vertical direction 91 by one stage of the CCD. Transfer (step S1 in FIG. 9). This state is shown in FIG.

  In the state of FIG. 3B, the noise charge 23 c existing in the final stage CCD of the high-speed transfer unit 60 at the time of FIG. 3A is transferred to the storage unit 6. Next, the second drive unit 82 does not transfer charges in the vertical direction 91 to the TDI pixel 1 of the light receiving unit 50, while the first drive unit 81 in the meantime, the vertical CCD 3 of the high-speed transfer unit 60. Then, charge is transferred in the vertical direction 91 (step S2 in FIG. 9). This state is shown in (c) and (d) of FIG.

By this driving operation, the noise charges 23 a and the noise charges 23 b accumulated in the high-speed transfer unit 60 are transferred to the storage unit 6, and all the noise charges 23 generated in the high-speed transfer unit 60 are stored in the storage unit 6. Accumulated.
At this time, the signal charge 22e in the final stage of the light receiving unit 50 at the time shown in FIG. 3A is transferred to the vertical CCD 3 one stage before the storage unit 6 in the vertical CCD 3 of the high-speed transfer unit 60. Retained. 3 is performed at high speed, that is, the frequency of the clock supplied from the first drive unit 81 to the high-speed transfer unit 60 is changed to a normal charge transfer operation. By raising (higher speed) than the frequency to be used, noise charges (smear) generated in the high-speed transfer unit 60 during this period are suppressed.

Next, when the high-speed transfer of the noise charge 23 shown in FIG. 3E is completed, the reset gate 8 (not shown in FIG. 3) is opened by the third drive unit 83 and the noise charge 23 is discharged. (Step S2 in FIG. 9).
Next, as shown in FIG. 3 (f), one stage of charge transfer is performed to the high-speed transfer unit 60, and the signal charge 22 e is transferred to the storage unit 6. Further, the accumulation control gate 7 is opened, and the signal charge 22e is transferred to the horizontal CCD 5 of the horizontal transfer section 70 as shown in FIG. 3G (step S3 in FIG. 9).
After completing the above high-speed transfer operation, the third drive unit 83 closes the accumulation control gate 7, drives the horizontal CCD 5, and reads the signal charge 22e from the output amplifier 9 as a signal for one imaging period in time series. .

  According to the driving method described above, the unnecessary photocharge 23 generated in the high-speed transfer unit 60 can be discharged independently of the transfer operation of the signal charge 22 during the blanking period of the imaging cycle. Mixing in the signal charge 22 can be prevented. Therefore, in a back-illuminated linear image sensor using a TDI-CCD, crosstalk can be reduced, and moving images can be captured without a mechanical shutter.

  In addition, since the high-speed transfer unit 60 is provided, and the high-speed transfer unit 60 is arranged at the boundary between the light incident region 19 and the light shielding region 19c, in the present embodiment, the boundary is defined as the boundary of the light receiving unit 50. There is no need to accurately align the edges. Therefore, the boundary is formed at an arbitrary position within the specified range of the high-speed transfer unit 60 in the arrangement direction of the vertical CCDs 3 constituting the high-speed transfer unit 60, and TDI- A back-illuminated linear image sensor using a CCD can be realized.

  In the driving method of the back-illuminated TDI-CCD image sensor 101 according to the first embodiment of the present invention shown in FIG. 3, the reset gate 8 is opened and the noise charge 23 is discharged at the time shown in FIG. However, the noise charge 23 may be discharged while the reset gate 8 is kept open during the period of (b) to (e) in FIG. According to this driving method, the noise charges 23a to 23c are not accumulated in the storage unit 6 during the high-speed transfer period. Therefore, it is not necessary to secure the saturation capacity of the storage unit 6, and there is an advantage that the circuit area can be reduced.

Embodiment 2. FIG.
FIG. 4 is an element plan view showing a circuit configuration of a back-illuminated TDI-CCD as an example of the back-illuminated image sensor 102 according to the second embodiment of the present invention. The back-illuminated TDI-CCD image sensor 102 according to the second embodiment includes a high-speed transfer unit 61 instead of the high-speed transfer unit 60 described in the first embodiment, and the reset gate 8 is deleted accordingly. The configuration is as follows. Other configurations are the same as those of the image sensor 101 in the first embodiment, and the description of the same components is omitted.

  The high-speed transfer unit 61 of the image sensor 102 includes a vertical CCD 24 provided for each column and an overflow unit 25 adjacent to the vertical CCD 24. The overflow unit 25 includes an overflow drain and a control gate (not shown).

  The operation, that is, the driving method of the back-illuminated linear image sensor 102 of the second embodiment configured as described above will be described with reference to FIGS. The driving method is executed for the light receiving unit 50, the high-speed transfer unit 61, the storage unit 6, the storage control gate 7, and the horizontal transfer unit 70 under the control of the drive unit 80.

FIG. 5 is a potential diagram of the back-illuminated TDI-CCD image sensor 102, and shows the potential pattern of the transfer channel 16 in time series.
Similarly to the light receiving unit 50 in the image sensor 101 of the first embodiment, also in the light receiving unit 50 in the image sensor 102, the signal charge 22 is integrated in the TDI pixel 1, and the signal charge 22 a in proportion to the number of stages of the TDI pixel 1. The amount of signal charge increases in the order of 22b,. Further, noise charges 23 are also accumulated in each vertical CCD 24 of the high-speed transfer unit 61.

  When one imaging cycle is completed as shown in FIG. 5A, next, at the time shown in FIG. 5B, the overflow gate of the overflow unit 25 is opened by the control of the first drive unit 81, and noise charge is generated. 23 is discharged to the overflow drain. Since the overflow unit 25 is provided for each vertical CCD 24, the noise charges 23a to 23c accumulated in the vertical CCD 24 constituting the high-speed transfer unit 61 are discharged simultaneously (step S11 in FIG. 10).

After the discharge of the noise charge 23 is completed, the first drive unit 81 closes the overflow gate. Then, the second drive unit 82 transfers the signal charge 22 integrated in the light receiving unit 50 by one stage in the vertical direction (step S12 in FIG. 10). Therefore, the signal charge 22e accumulated in the final stage of the light receiving unit 50 is transferred to the first stage of the high-speed transfer unit 61. The state after transfer is shown in FIG.
Next, the drive unit 80 does not transfer charges to the light receiving unit 50, and sequentially transfers charges only to the high-speed transfer unit 61 in the vertical direction (step S13 in FIG. 10). This state is shown in (d) to (f) of FIG. The signal charge 22e is transferred to the storage unit 6 by performing a series of transfer operations of (d) to (f) in FIG.

  Next, as shown in FIG. 5G, the accumulation control gate 7 is opened, and the signal charge 22e is transferred to the horizontal CCD 5 (step S13 in FIG. 10). After completing the above transfer operation, the control gate 7 is closed, the horizontal CCD 5 is driven, and the signal from the output amplifier 9 is read out for one imaging period in time series.

  As described above, according to the back-illuminated linear image sensor 102 in the second embodiment, unnecessary noise charges generated in the high-speed transfer unit 61 are provided by providing the overflow unit 25 in the vertical CCD 24 for each column of the high-speed transfer unit 61. 23 can be discharged at the overflow section 25 in each vertical CCD 24 at the same time. By this operation, as in the image sensor 101 in the first embodiment, the noise charge 23 is discharged independently of the transfer operation of the signal charge 22 and mixed into the signal charge 22 during the blanking period of the imaging cycle. There is no. Therefore, in a back-illuminated linear image sensor using a TDI-CCD, crosstalk can be reduced, and moving images can be captured without a mechanical shutter.

  Similarly to the high-speed transfer unit 60 in the image sensor 101 according to the first embodiment, the high-speed transfer unit 61 is disposed at the boundary between the light incident region 19 and the light shielding region 19c. It is not necessary to accurately align the boundary portion with the end of the light receiving unit 50. Therefore, the boundary portion is formed at an arbitrary position within the specified range of the high-speed transfer section 61 in the arrangement direction of the vertical CCDs 24 constituting the high-speed transfer section 61, and TDI- A back-illuminated linear image sensor using a CCD can be realized.

  In the image sensor 102 according to the second embodiment, since the overflow unit 25 is provided in the vertical CCD 24 for each column of the high-speed transfer unit 61, the width of the vertical CCD 24 is smaller than that of the image sensor 101 according to the first embodiment. . As the width of the vertical CCD decreases, the charge transfer capacity decreases. However, since the high-speed transfer unit 61 is not intended for light detection, the charge transfer capacity can be secured by extending the length of each vertical CCD 24 per stage.

Embodiment 3 FIG.
FIG. 6 is a plan view showing a circuit configuration of the image sensor 103 of the back-thinned TDI-CCD according to the third embodiment of the present invention. The image sensor 103 has a configuration in which a light shielding film 27 made of a light shielding material is further added to the image sensor 101 in the first embodiment. Other configurations are the same as the configuration of the image sensor 101. Therefore, only the light shielding film 27 will be described below. In addition, the code | symbol 26 shown in FIG. FIG. 7 is a device cross-sectional view schematically showing the cross-sectional structure of the BB ′ portion shown in FIG.
Further, in the third embodiment, as described above, the image sensor 101 of the first embodiment is provided with the light shielding film 27. Of course, the image sensor 102 of the second embodiment is provided with the light shielding film 27. It is also possible to adopt.

As shown in FIGS. 6 and 7, the light shielding film 27 is formed so as to cover the light incident region 19 thinned from the back surface 12 a side of the substrate by Si anisotropic etching and the light shielding region 19 c other than the light incident region 19. After that, it is removed by etching so as to form an opening 19b only in a part of the bottom surface 19a of the recess of the light incident region 19. As shown in FIGS. 6 and 7, at least the light receiving unit 50 is positioned corresponding to a region where the light shielding film 27 does not exist on the bottom surface 19 a.
The operation of the image sensor 103 according to the third embodiment is the same as the driving method of the image sensor 101 according to the first embodiment described above, and thus the description thereof is omitted here.

Hereinafter, a method of manufacturing the back-illuminated TDI-CCD image sensor 103 according to Embodiment 3 will be described with reference to FIG.
As a substrate wafer, as shown in FIG. 8A, an SOI substrate 15 made of bulk Si12, BOX oxide film 13, and Si epitaxial layer 14 is used. Using a standard Si process, as shown in FIG. 8B, a CCD having a transfer channel 16 and a transfer gate 17 on the surface side of the SOI substrate 15, that is, a light receiving unit 50, a high-speed transfer unit 60, A horizontal transfer unit 70 and the like are formed.

In FIG. 8C, the back surface 12a of the bulk Si 12 is mechanically polished to thin the bulk Si 12 to a desired film thickness. Thereafter, as shown in FIG. 8D, an etching mask 21 is formed on the back surface 12a. The etching mask 21 is formed so as to be positioned corresponding to the storage unit 6, the storage control gate 7, the reset gate 8, and the horizontal transfer unit 70 described above.
In FIG. 8D and subsequent figures, the back surface 12a is shown in the upper part of the drawing. At this time, since the substrate back surface 12 a is flat, a normal Si process can be used for patterning the etching mask 21.

  Next, in (e) of FIG. 8, Si anisotropic etching is performed from the back surface 12a side by wet etching using an etchant such as KOH or TMAH to form a light incident region 19 having a light step structure. At this time, since the BOX oxide film 13 acts as an etching stopper, the Si film thickness under the light receiving portion 50 can be formed uniformly.

  Next, as shown in FIG. 8F, a light-shielding material having photosensitivity is applied to the light incident region 19 on the back surface 12a by using a spray coating method. In the standard Si process, a spin coating method is used to apply a photosensitive material such as a resist. However, in the case of a high step structure such as the light incident region 19, the spin coating method causes the resist to accumulate only in the recessed portion. The problem that it is not uniformly applied arises. Therefore, the spray coating method is used as described above. In the third embodiment, since the light shielding material is sprayed and applied by the spray gun 28, the light shielding material can be uniformly applied even in a high step structure.

  Next, in FIG. 8G, the light shielding film 27 is exposed using an exposure machine such as a double-sided aligner or a double-sided stepper, and subsequently developed to perform patterning. In general, in exposure for a high step structure, it is difficult to focus on both a high step portion and a low step portion when a mask pattern is projected onto a wafer. However, in the third embodiment, since the pattern to be opened is laid out so as to be located only on the bottom surface of the high step structure, focusing is easy and a desired pattern can be projected.

  At this time, the mask alignment is performed with respect to the circuit pattern of the CCD on the substrate surface side, that is, with reference to the circuit pattern. For this reason, even if the circuit pattern on the front surface and the edges 10 and 11 of the light incident region 19 on the back surface side are not formed in parallel due to an angle shift such as orientation flat due to manufacturing tolerances as shown in FIG. The edge 26 of the light shielding film 27 can be formed in parallel to the circuit pattern. The transmittance of the light shielding film 27 is desirably as small as possible. Therefore, the thickness of the light shielding film 27 needs to be formed to a thickness of several μm or more.

  In this case, it is difficult to finish the shape of the edge 26 of the light shielding film 27 into a clean straight line shape, and in fact, it tends to be a curved shape having unevenness as shown in FIG. However, as in the case of the image sensor 101 in the first embodiment, the high-speed transfer unit 60 is provided, so that the boundary between the light incident region 19 and the light shielding region 19c is accurately aligned with the end of the light receiving unit 50. do not have to. That is, the boundary only needs to be located within the specified range of the high-speed transfer unit 60 in the arrangement direction of the vertical CCDs 3 in the high-speed transfer unit 60, and by doing so, back-illuminated TDI− excellent in sensitivity uniformity between pixels. A CCD can be realized.

The image sensor 103 according to Embodiment 3 described above has the following advantages.
That is, by providing the light shielding film 27, it is possible to improve the light shielding properties for the storage unit 6, the accumulation control gate 7, the reset gate 8, and the horizontal transfer unit 70 compared to the case where the light shielding film 27 is not provided.
Further, since the light shielding film 27 does not require a processing step such as etching and can be patterned only by an exposure step, the manufacturing yield is improved and only the bottom surface 19a of the light incident region 19 needs to be patterned. Therefore, the exposure apparatus can be easily focused on a structure having a relatively large step difference, such as the light incident area 19, and patterning accuracy is improved.

  In the above-described examples of the first to third embodiments, the case where the SOI substrate 15 is used as the Si substrate has been described. However, the same applies to a back-illuminated sensor formed using another wafer such as an epi wafer. is there.

1 is an element plan view of a back-illuminated TDI-CCD according to Embodiment 1 of the present invention. FIG. 2 is an element cross-sectional view of a back-illuminated TDI-CCD in the A-A ′ portion shown in FIG. 1. It is a potential diagram for demonstrating the drive method of the back-illuminated type TDI-CCD shown in FIG. It is an element top view of back incidence type TDI-CCD by Embodiment 2 of the present invention. FIG. 5 is a potential diagram for explaining a driving method of the back-illuminated TDI-CCD shown in FIG. 4. It is an element top view of the back-illuminated TDI-CCD according to the third embodiment of the present invention. FIG. 7 is an element cross-sectional view of a back-illuminated TDI-CCD at a B-B ′ portion shown in FIG. 6. It is a figure explaining the manufacturing method of the back-illuminated TDI-CCD shown in FIG. 2 is a flowchart for explaining a method of driving the back-illuminated TDI-CCD shown in FIG. 5 is a flowchart for explaining a method of driving the back-illuminated TDI-CCD shown in FIG.

Explanation of symbols

1 TDI pixel, 3 vertical CCD, 5 horizontal CCD, 6 storage unit,
7 Storage control gate, 8 Reset gate, 12 Bulk Si,
12a back surface, 16 transfer channel, 17 gate electrode, 19 light incident region,
19c light shielding area, 22 signal charge, 23 noise charge, 24 vertical CCD,
25 overflow part, 27 light shielding film,
50 light receiving unit, 60, 61 high-speed transfer unit, 70 horizontal transfer unit, 80 drive unit,
101-103 Back-illuminated linear image sensor.

Claims (7)

Charge transfer device implementing the back-illuminated type TDI process for transmitting a signal charge generated in the charge transfer device by light incident from the light incident region formed on the back surface side of the semiconductor substrate which are arrayed in rows on the surface A linear image sensor,
Having a first charge transfer element arranged in a first direction along the surface of the semiconductor substrate opposite to the light incident region, the signal charge generated and a light receiving portion for transferring to the first direction ,
A horizontal transfer having a second charge transfer element arranged in a second direction along the surface facing a light shielding region other than the light incident region on the back surface side, and transferring the signal charge in the second direction; And
A third charge transfer element arranged between the light receiving portion and the horizontal transfer portion along the surface and arranged in the first direction at a boundary portion between the light incident region and the light shielding region; wherein said third charge transfer device, the light of the signal charges supplied from the light receiving unit is incident from only One light input region transferred to the first direction is incident on the third charge transfer device noise A high-speed transfer unit that generates electric charge;
High-speed transfer that causes the high-speed transfer unit to perform charge transfer without performing charge transfer to the light-receiving unit, discharges only the noise charge, and transfers the signal charge transferred from the light-receiving unit to the horizontal transfer unit A drive unit that performs transfer operation control; and
A back-illuminated linear image sensor.   The portion that discharges the noise charge, is disposed between the high-speed transfer unit and the horizontal transfer unit along the surface so as to face the light shielding region, and is controlled by the high-speed transfer operation control of the drive unit. The back-illuminated type according to claim 1, further comprising an accumulation control unit that accumulates and discharges the noise charge transferred from the high-speed transfer unit and transfers the signal charge transferred from the light receiving unit to the horizontal transfer unit. Linear image sensor. The high-speed transfer unit includes an overflow unit that is disposed in each of the third charge transfer elements included in the high-speed transfer unit and discharges the noise charge generated by the third charge transfer element. The back-illuminated linear image sensor according to claim 1, wherein the overflow charge is discharged after the end of one imaging cycle.   The light incident region is a concave region formed on the back surface of the semiconductor substrate, and the back-illuminated linear image sensor has a light-shielding film that has an opening for light incident on the bottom surface of the recess and covers the back surface. The back-illuminated linear image sensor according to any one of claims 1 to 3, further comprising: A driving method of a back-illuminated linear image sensor according to claim 1 or 2,
Performs light receiving portion and the charge transfer only one stage for all of the first charge transfer device and a third charge transfer devices arranged in the high-speed transfer section to the first direction, the first final stage in the light receiving portion The signal charge existing in the charge transfer element is transferred to the high-speed transfer unit,
After transferring the signal charge, the charge transfer to the light receiving unit is stopped, while the charge transfer is performed only to the high-speed transfer unit and the noise charge generated by the third charge transfer element of the high-speed transfer unit is generated. Discharge
After discharging the noise charge, the signal charge transferred from the light receiving unit to the high-speed transfer unit is transferred to a horizontal transfer unit.
A driving method for a back-illuminated linear image sensor. A driving method of a back-illuminated linear image sensor according to claim 3,
After the end of one imaging cycle, the generated noise charge is discharged to all the third charge transfer elements of the high-speed transfer unit,
After discharge of the noise charge, all of the first charge transfer device provided in the light receiving unit, and performs a charge transfer only one stage for all of the third charge transfer device provided in the high-speed transfer section to the first direction , Transfer the signal charge present in the first charge transfer element of the final stage in the light receiving unit to the high-speed transfer unit,
While stopping the charge transfer to the light receiving unit, transfer the charge to only the high-speed transfer unit and transfer the signal charge to the horizontal transfer unit,
A driving method for a back-illuminated linear image sensor. A method for manufacturing a back-illuminated linear image sensor according to any one of claims 1 to 4,
A concave light incident region is formed on the back surface of the semiconductor substrate on which the light receiving unit, the high-speed transfer unit, and the horizontal transfer unit are formed, wherein the light receiving unit is positioned corresponding to the light incident region, and the high-speed transfer unit Is located at the boundary between the light incident region and the light shielding region other than the light incident region,
Forming a light shielding film covering the light incident region and the light shielding region;
When forming the opening by removing the light shielding film in the light incident region, the alignment of the opening forming mask is aligned with the pattern of the charge transfer element in the light receiving unit or the high-speed transfer unit.
A method of manufacturing a back-illuminated linear image sensor.

Images