JPS6114748A - Color image sensor - Google Patents

Abstract

PURPOSE:To improve the sensitivity of a photodiode for a red color and to eliminate a crosstalk by inserting an amorphous germanium layer between a blocking diode and the photodiode, thereby preventing the leakage current of the blocking diode. CONSTITUTION:Chromium is deposited as an individual electrode 22 on an insulating substrate 21, a phosphorus-doped n type a-Si layer 41, an i-type a-Si layer 42 and a boron-doped p type a-Si layer 43 are accumulated to form a blocking diode. An amorphous germanium 44 is accumulated as a light absorbing layer. Boron-doped p type and i-type a-SiGe layers 45, 46 are laminated, and a phosphorus-doped n type a-Si layer 47 is accumulated to form a photodiode. A protective layer 48, a transparent conductive film 24 and a light shielding film 25 are formed. Red, green and blue dies are thermally diffused in a polymer 49 to form a color filter 50.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a color image sensor using amorphous silicon.

(Prior Art and its Problems) Recently, contact image sensors have been actively developed in order to miniaturize document reading devices such as facsimiles. This is a photoelectric conversion device that incorporates a long-dimensional image sensor whose length is the same as the width of the document. Traditional COD
A photoelectric conversion system having a TC image sensor or a MOS type TC image sensor requires a reduction lens system to project the original onto the image sensor.

In order to secure the optical path length, there is a large space inside the device! This has the disadvantage that it is difficult to miniaturize the device.

Totoro, the above-mentioned contact type image sensor does not require this reduction lens system, and the image sensor and document are in close contact with each other.
Therefore, the device can be significantly downsized.

This contact image sensor uses an arsenic-selenium-tellurium amorphous semiconductor (T, Tukada) as a photoelectric conversion material.
et al., IEDM, 546 (1977)) + cadmium sulfide (Komiya et al., Confucian Journal.

Vo I J-64C, 8117, l 9'81 7
), hydrogenated amorphous silicon (hereinafter abbreviated as a-8i)
(S.Kanek.

al., IEDM, 328 (1982)) has been developed. All of these materials are known to meet the requirements for realizing large-scale image sensors, such as being able to form a uniform thin film over a large area and having high sensitivity to visible light. Among these, a-8i is a non-polluting material whose constituent elements are silicon and hydrogen.
(It is a stable material compared to other amorphous semiconductors, as it has a heat resistance of about 1'C, has high photosensitivity in a wide range of visible light, and is a thin film with high resistivity, so it can be used for storage type drive. It is considered to be a suitable material for color image sensors, as it has the first-rate feature of enabling high speed and high resolution.

When applying the a-8+ to a contact image sensor,
A system in which the a-8i optical sensor is driven by Si MOS PET and an X switch is normally used. Figure 1 a-
In Figure 1, which shows the equivalent circuit of an 8i image sensor, an 11-digit a-8i optical sensor, 12tJ: Si MO8
FFliT switch array 13. It is an 81 IC chip with a built-in scanning circuit wJ14. This Si MOS
The PPT switch 13 and the individual electrodes 15 are connected by means such as bonding. The a-8i optical sensor is a photodiode type with a fast optical response speed, and can realize an image sensor that can read documents at high speed.

The photocurrent generated by the light irradiated to the a-8i photodiode 11 erases the charge in the diode, and after a certain scanning time (accumulation time), the MO8F'ET switch 1 is turned on again.
3 turns ON, a charging current flowing from an external terminal is detected, and an optical signal is read out from K. In this signal detection method.

The amount of optical signal charge is proportional to the product of scanning time and light receiving area. On the other hand, what is the amount of noise included in the read signal?

This is because the dark current of the photodiode is very small.

It is dominated by the switching noise of the general MO8FET13. Therefore, since the noise component is considered to be constant regardless of the scanning time or the incident light intensity iK, the faster the image sensor operates or the higher the resolution, the smaller the optical signal becomes and the 8/N ratio decreases.

In order to colorize a long image sensor using this amorphous silicon, a conventionally reported example is to use a monochromatic one-dimensional sensor array as a light receiving element, and to colorize a long image sensor using red (R) and green (G). , blue (B) are used to read one line of an i draft, and then switch to KM. There is a way to colorize the image by switching from K to K. .

However, the disadvantage of this light source switching method is that the storage time is 1/3 shorter than that of the color method, resulting in a decrease in reading speed and a decrease in the S/N ratio. This is an essential drawback of the light source switching method. Therefore, rather than switching the light source, it is more advantageous to use white light as the illumination light and to make each optical sensor sensitive to each color.

A photosensor with color sensitivity can be realized by providing a color filter on the light-receiving surface of a photodiode.The structure of a color image sensor using a 0-color filter is as follows.
Generally, a-8i23 is deposited on the individual electrodes 22 as shown in FIGS. 2(al) and (b)K.

A transparent conductive film 24 is provided on the surface of the photodiode, and a color filter 26 is provided on the light incident surface of the photodiode. Here, 25 is a light-shielding film provided to improve resolution, but it is not essential. The same can be said of the sensor of the present invention which will be described later.

However, in the above case, in order to obtain the same resolution as compared to a monochromatic image sensor, it is necessary to increase the density of the photosensor by the amount of color that can be differentiated. For example, in a full-color image sensor, it is necessary to create three or more photodiodes of a size corresponding to one seven-photo sensor of a monochrome image sensor, which reduces the light-receiving area of each element. As a result, the amount of optical signals decreases, 8
This results in a decrease in the /N ratio, making high-speed operation difficult. Also, as mentioned above, there is a one-to-one correspondence between photosensors and MOS FET switches, so colorization requires more than three times as many photosensors.

The number of connection points such as bonding increases accordingly, and the number of IC chips also increases. This reduces the reliability of the contact type image sensor and increases cost. moreover.

The higher the resolution, the more precise the alignment between the photodiode and filter, which makes the process more difficult.

As a structure to eliminate these drawbacks, the same applicant as the present applicant proposed a color sensor as shown in FIGS. 3(a) and 3(b). The figure requires full color.

For example, a photosensor sensitive to red, green, and blue light is 1
They are arranged in one row on the individual electrodes 22, respectively. Therefore, when compared with Fig. 2, it is clear that even though the area of one photodiode is the same, it has three times the resolution. I can do it. Furthermore, even if color separation of four or more colors is performed using complementary colors, there is an advantage that the amount of optical signal charge does not decrease due to an increase in the number of sensors per one reading pixel. Furthermore, since the filters can be attached in a striped pattern, alignment is easy even when the resolution is high fK.

By the way, 1. For example, in FIG. 3, in the vertical direction, R, G,
BK: There are three corresponding common electrodes, one in the horizontal direction
728 lines (A4 size, pixel density 8 elements/,)
There are several individual electrodes. Since the photosensor is formed at the intersection of the common electrode and the individual electrodes, it can be seen that it is a 3×1728 matrix circuit in terms of one circuit configuration. Suppose that we are currently reading out an ordinary signal. At this time, the photodiodes in the portions corresponding to G and H are in the accumulation state 11.
, and each photodiode must be held at a respective cumulative voltage corresponding to the amount of charge being charged. However, one individual electrode is for each photodiode 3 of R.
Since the voltage corresponds to 2, an element is required to compensate for this difference in voltage. This can be achieved by using a reverse bias condition of the diode. Hereinafter, this diode 31 will be referred to as a blocking diode. In order to function in this way, K must reduce the leakage current in the reverse direction of the blocking diode. When driving the photodiode in a matrix using a blocking diode in this way, as shown in the cross-sectional view in Figure 3, A method has been proposed in which a blocking diode 31 and a photodiode 32 are stacked. By stacking two diodes in this manner, there is no need for the two-layer wiring used in conventional matrix circuits, and a color sensor can be realized with a simple structure.

However, in a structure in which a blocking diode based on amorphous silicon and a photodiode are simply stacked, the component of the light incident on the light receiving surface of the photodiode that is not absorbed by the photodiode reaches the blocking diode section K. However, some of them generate electron-hole pairs, causing leakage current to flow through the blocking diode. In particular, red light with a wavelength of about 650 nm has relatively low light absorption in the A-8I, and the A-8I photodiode has low sensitivity to red. It has become clear that such red light reaches the blocking diode, causing significant leakage current in the blocking diode, resulting in crosstalk between photosensors.

(Object of the Invention) The present invention eliminates these drawbacks, and provides a color image sensor that has high speed, high resolution, has low crosstalk, improves the sensitivity of the photodiode to red, and can be manufactured in a simple process. There is K to provide.

(Structure of the Invention) According to the present invention, a plurality of metal electrodes divided into bits are provided on an insulating substrate, and a blocking diode based on an amorphous semiconductor having silicon as a constituent element is formed on the skeleton electrode. In a color image sensor in which at least two rows of photodiodes are stacked in this order and a color filter is provided on the surface of the stacked structure, the photodiode is connected to the bit where light of a wavelength that at least passes through the photodiode is incident. A color image sensor is obtained, in which part or all of the photodiode contains germanium, and an amorphous germanilum or an amorphous semiconductor layer containing germanium is inserted between the blocking diode and the photodiode.

(Detailed description of the configuration) The present invention solves the problems of the prior art by adopting the above-mentioned configuration. In a structure in which a photodiode and a blocking diode are stacked, what is the cause of leakage current in the blocking diode?

This is due to imperfection in light absorption in the photodiode section.

The wavelength dependence of the light absorption coefficient of a-8+ is 105C+1-1 or more for light as short as 5QOnm, and it is considered that all incident light of such a wavelength is absorbed by the photodiode section. However, since the light absorption coefficient around 650 nm is about 10'Cm', such light passes through the photodiode and reaches the blocking diode.
Some generate photocurrents in blocking diodes. Looking at it from another perspective, it can be said that the a-81 photodiode has low sensitivity to red. This is the cause of crosstalk. Furthermore, the absorption coefficient of a-8i is small for light in the long wavelength infrared region, and even if the light reaches the blocking diode, no electron-hole pairs are generated, so there is no problem. Therefore, in order to prevent the leakage current of the blocking diode, K should improve the sensitivity of the photodiode to red light and allow the photodiode to absorb all the light in the vicinity of fi50 nm.

In order to increase absorption, it is generally sufficient to increase the thickness of the film, but in reality there are problems such as constraints on deposition time, cost, and effects on diode characteristics, so this cannot be a fundamental solution to the problem. Therefore, it is thought that it is better to form the photodiode with a material that has a smaller optical bandgap than a-8i and has a large optical absorption coefficient even for light with a wavelength of about +650 nm.

However, to use it as a photodiode, Vi needs to be a material that can form a high-quality junction with low leakage current and has high photoconductivity. In short, the semiconductor must have a small optical bandgap and be of high quality. As a material that satisfies these conditions, amorphous silicon germanium (hereinafter referred to as a-
8iGe) (for example, M, C, Cr
Tella et al.

J, Electrochem, Soc, , 1
2.2850 (1982)).

a-8i Ge Pk, which can be formed by the same glow discharge decomposition method as a-8i. Moreover, when the proportion of germanium in a-8iGe is changed, the optical band gap also changes.

The larger the proportion of germanium, the more it absorbs light with longer wavelengths.

However, in general, the higher the proportion of germanium, the more
The quality of the iGe film deteriorates. For this reason, when germanium is mixed into a photodiode to increase the light sensitivity at long wavelengths, the reverse leakage current of the diode increases (
7. Therefore, germanium is mixed into the photodiode to the extent that the diode characteristics do not deteriorate to increase the sensitivity in the long wavelength region. In order to prevent the light that cannot be completely absorbed by the photodiode from reaching the blocking diode, an a-8iGe or amorphous material containing a large amount of germanium, which has a large absorption coefficient even for long wavelength light, is placed between the blocking diode and the photodiode. Germanium is inserted as a light absorption layer.

(7) By mixing germanium into a part of the photodiode and providing a light absorption layer.

The long wavelength sensitivity of the photodiode is increased without deteriorating the diode characteristics, and the light shielding effect is also perfected. Another advantage of the present invention is that restrictions on device design, such as the thickness of each layer and the amount of germanium mixed, are relaxed, and each element can be manufactured under optimal conditions.

The structure of the photodiode is p-1-n type, metal or TTO.
M using a Sittke junction type with or a thin insulating layer
There is an IS type. The present invention works effectively with any structure. For example, in the p-4-n type, if the light-receiving side is the p layer, then the n
layer only, i layer and n layer, or all a
By replacing it with -8IGe, the photodiode section can absorb light with a wavelength of around 650 nm. Similarly, by using a-8iGe in part or in whole in other structures, it is possible to prevent the leakage current of the blocking diode from occurring.

(Example) Examples of the present invention will be described in detail below with reference to the drawings. FIG. 4 shows a detailed cross-sectional view of the image sensor according to the first embodiment of the present invention. Individual electrodes 2 are mounted on an insulating substrate 21 such as a glass or ceramic substrate.
2 and 1. Chromium is deposited to a thickness of 500 .ANG. and etched to a width of about 100 .mu.m in the case of an image sensor having a resolution of the necessary linear density of 1, for example, 8 lines/. Next, N-type A-8i 41 doped with 0.2% phosphorus by silane glow discharge decomposition method, etc., was made to a thickness of 0.2 am, I-type A-8i42 was made to a thickness of 0.6 μm, and boron was doped to a thickness of 0.2 μm. Deposit 0.2 μm thick doped p-type A-8I43 to form a blocking diode. Furthermore, as a light absorption layer, amorphous germanium 44 was deposited to a thickness of 0 by glow discharge decomposition of germanium (GeH4) gas.
Deposit 2 μm. The composition ratio of silicon to germanium in this film was about 0.7:0.3. Besides, K.

N-type 45 doped with 0.2% boron by glow discharge decomposition of a mixed gas of silane and germane. i type 46, (
D a-81()e with thicknesses of 02 μm and 0.6 μm, respectively.
Finally, n-type a-Si47 doped with 0.2% phosphorous by glow discharge decomposition of silane is deposited to a thickness of 0.5m.
02 μm is deposited to form a photodiode. Next, the photodiode, light absorption layer, and blocking diode were simultaneously etched into an island shape using a conventional dry etching method.

Separate each element. The area of this island is 1, for example, 8 dots/
For an image sensor with a density of 100μm x 140μ
It was m. Thereafter, in order to prevent the generation of IJ current on the etched surface, a protective layer of SiO□48 was formed to a thickness of 08 μm using sputtering or the like.

After making a contact hole by etching, indium tin oxide is deposited to a thickness of 0.06 mm as a transparent conductive film 24.
Chromium is vapor-deposited to a thickness of 01 μm as the light-shielding film 25, and processed into stripes as a common electrode by etching. Finally, after spin-coding the polymer 49, a color filter 50 is formed by thermally diffusing red, green, and blue dyes into the polymer. The size of one stripe of this color filter is approximately 180μm x 22
It was 0m.

l and p with color filters attached by the above process @
7 Using VCa-8ice, an array of elements in which a diode/light absorption layer/blocking diode are stacked is completed. In this embodiment, Ge is mixed in the i-layer and the n-layer, but a-8iGe may be used in all the layers including the p-layer. Also, in this embodiment, when viewed from the light-receiving surface side, an n-1-p structure is used for the photodiode and a p-1-n structure is used for the blocking diode, but conversely, the photodiode Kp-i-n , n-i to the blocking diode
The present invention is also effective using a combination of -p structures.

In addition, in this embodiment, the blocking diode is Kp type, n
I am using an A-8i model.

p-type, i-type, n-type a-8iC, 8iN or microcrystalline S
It is known that a diode can be formed by using blocking diode 1 partially or completely, but the present invention is also effective even if such a blocking diode is used. Also, in the photodiode, instead of the n-type a-8i on the light receiving surface, a
The present invention is also effective using materials such as -8iC, a-8iN, or microcrystalline Sr. Furthermore, the present invention also applies if a blocking diode or photodiode is formed using metals such as Cr, Pt, Pd, Au, or indium tin oxide as one individual electrode or an upper transparent electrode, and utilizes a brush contact with a-8i. It is valid.

Furthermore, although the light absorption layer needs to have low resistance, since amorphous germanium is a low resistance film, it can be used without doping as in this embodiment. When using a-8iGe, it is necessary to dope it with phosphorus or boron to lower the resistance.

FIG. 5 is a sectional view of a second embodiment of the invention.

In this embodiment, n-type 51. n-type 52 * n-type 53-
A blocking diode is formed by sequentially stacking 8i, and then a G doped with 1 PH3 is used as a light absorption layer 54.
e N-type a-8iGe with a thickness of 04 μm is formed by glow discharge decomposition of a mixed gas of H4 and 5iH4. The composition ratio of silicon and germanium in this film is approximately 0.4:0.
．． It is 6. Furthermore, n-type 55 doped with 0.51 phosphorus
and i-type 56a-8iGe with a thickness of 0.2 μm and 1
A photodiode was formed by depositing .mu.m of the film and making use of a mechanical contact with ITO.

The other structures are the same as in the first embodiment.

(Effects of the Invention) Using the laminated array of photodiodes and blocking diodes according to the embodiment of the present invention, a MOS IC including a scanning circuit and a switching element is installed on a substrate, and a MOS FET switch and a diode are bonded. When connected and operated as an image sensor, the signal output was tripled compared to the conventional image sensor with the same resolution as shown in Figure 1.
A high-speed, high-resolution color image sensor without crosstalk was obtained, confirming the effectiveness of the present invention.

Note that in this example, all pins) K: a-8iGe
However, the a-
The present invention is also effective using 8iGe. In addition, for the purpose of identifying 1% constant colors (for example, blue and red) [7 sensors, 2
A column diode array is sufficient.

[Brief explanation of drawings]

Figure 1 is an equivalent circuit diagram of a conventional image sensor; Figure 2 (a) and (bl) are cross-sectional views and plan views of a conventional color image sensor; Figure 3 (b) is a conventional color image sensor; A cross-sectional view and a plan view of an image sensor. Fig. 4 is a cross-sectional view of the first embodiment according to the present invention, and Fig. 5 is a cross-sectional view of the second embodiment according to the present invention. In the figures. 11...・Optical sensor, 12...St IC
Chip. 13...MOS FE'I' switch, 14...
...Scanning circuit. 15...Individual electrode, 16...Common electrode, 21...Insulating substrate 122...Individual electrode, 23...Photodiode, 24... Transparent electrode, 25... light shielding film. 26...Color filter, 31...Blocking diode, 32- Photodiode, 41...
N type a-8t, 42 ・= ・L type a-8i, 43
--p-type a-8i, 44---a-Ge, 4
5---・p-type a-8jGe. 46---i type a-8i, 4'7---nfJa-
8i, 48...5i02, 49...polymer, 50...color filter, 51-p type a-8
i, 52==-i type a-8i53--...n type a-
8i, 54.55++++n type a-84Ge, 5
fi・-・I type a-8iGe, 57・・・-・5i
02 respectively.

Claims (1)

[Claims] A structure in which a plurality of metal electrodes divided into bits are provided on an insulating substrate, and a blocking diode and a photodiode based on an amorphous semiconductor whose main constituent element is silicon are laminated in this order on the electrode. In a color image sensor in which at least two rows or more are provided and a color filter is provided on the surface of the laminated structure, a part or all of the photodiode contains germanium for a bit on which light of a wavelength that is transmitted through at least the photodiode is incident. , and an amorphous germanium or an amorphous semiconductor layer containing germanium is inserted between the blocking diode and the photodiode.