JPH0570946B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (Industrial Application Field) The present invention relates to a semiconductor photodetection device that detects light irradiation.

(Prior Art) A semiconductor photodetection device in which a plurality of photodiodes are arranged in an array on the same substrate is used for position detection, spectroscopic measurement, and the like.

In such a photodiode array type semiconductor photodetecting device, various problems arise when the degree of integration of the photodiodes is increased in order to improve the resolution of the incident position of incident light.

The first problem is optical crosstalk, in which light incident between adjacent photodiodes causes mutual interference between the elements. Optical crosstalk will be explained with reference to FIG.

FIG. 5 is a cross-sectional view of the device illustrating optical crosstalk in a photodiode array type semiconductor photodetection device.

Optical crosstalk occurs when light with a small absorption coefficient reaches a deep part of a semiconductor device far from the PN junction, generates electron-hole pairs inside, and these carriers diffuse into adjacent parts of the same array. This occurs by reaching the photodiode that

For example, this is the case when carriers generated in a deep part of the N layer 1 below the P ⁺ region 15 reach the P ⁺ region 16.

There is a physical crosstalk called secondary blooming. This physical crosstalk will be explained with reference to FIG.

Physical crosstalk occurs when the charge accumulated in the depletion layer shown by the broken line in the figure becomes saturated due to strong light irradiation, and then diffuses within the device and reaches an adjacent photodiode.

This is the case, for example, when carriers generated in the depletion layer formed in the N layer 1 below the P ⁺ region 15 reach the P ⁺ region 16.

These crosstalks blur the position boundaries at the position sensor and blur the distinction between two adjacent signal peaks at the analysis sensor.

Furthermore, in recent years, image sensors that combine a photodiode array and a self-scanning circuit (shift register) for signal reading have been widely used.

FIG. 7 is a sectional view showing one element of such an image sensor.

P channel using such N substrate 1
Figure 8 shows an equivalent circuit of an image sensor with a MOSFET structure.

In the P channel MOSFET structure using the N substrate 1, the PN junction in the region of the source 2 is used as a light receiving surface.

When a negative pulse voltage is applied to the gate electrode 4, a P channel is generated on the silicon surface under the gate electrode 4, and at this time, charge is supplied from the drain 3,
This diode is charged by making the source 2 diffusion junction equal to the drain 3 voltage.

When the voltage of the gate electrode 4 is turned off and the channel is closed, the potential of the source 2 (accumulated charge) is maintained as it is.

When the carrier is excited by the incident light in this state, the accumulated charge is discharged to this carrier, and the source 2
The potential of decreases. Next, when a scanning pulse is applied again via the gate electrode 4, the charged charges relative to the discharged charges flow into the source 2 and are taken out to the external circuit. Repeat this operation below.

In such a photodiode array,
The potential of each element in the array is always different. When the potential of a photodiode at a certain point increases due to the scanning pulse of the shift register, a phenomenon occurs in which current flows into an adjacent photodiode with a lower potential. This is electrical crosstalk and is particularly likely to occur when the distance between the photodiodes becomes short or when a high resistance substrate is used.
Electrical crosstalk makes accurate optical signal measurements and weak light measurements difficult.

These optical, physical, and electrical crosstalks will be collectively referred to as crosstalk hereinafter.

To deal with such crosstalk, a configuration can be considered in which the photodiodes are separated by a P ⁺ layer 5 formed by diffusion, as shown in FIG.

In this method, although the elements are separated for the time being, all carriers generated by light incident near the separation region 5 are absorbed by the P ⁺ type separation region 5, resulting in a reduction in signal amount. Also separation area
When the P ⁺ layer 5 is provided close to the photodiode, the interaction between the isolation region P ⁺ layer 5 and the P layer 15 of the photodiode becomes stronger, so that the accumulated signal charges are transferred to the isolation region P ⁺ layer 5. electrical crosstalk increases. The depth of the isolation region must be at least 5 μm, but when it is formed by diffusion, not only the depthwise diffusion but also the same degree of lateral diffusion occur simultaneously. Therefore, it is difficult to control the width of the isolation region P ⁺ layer 5 to 10 μm or less. Furthermore, when the depletion layer that has spread from the P layer 15 on the light-receiving surface reaches the isolation region P ⁺ layer 5, the withstand voltage drops significantly, so the distance between the light-receiving surface and the isolation region must be at least about 20 μm. As a result, the spacing between adjacent photodiodes is greater than 50 μm, which reduces the resolution and resolution performance limits of the photodiode array.

As mentioned above, between adjacent photodiodes
Although forming a P ⁺ isolation region is effective in preventing crosstalk, it increases leakage current, lowers withstand pressure, and causes signal leakage, which increases the ambiguity of the signal and adversely affects the characteristics of the device. decreases.

Furthermore, as shown in FIGS. 10 and 11, a configuration in which isolation regions made of an insulator are formed between photodiodes can be considered.

Carriers generated in the photodiode on the left side of FIG. 10 are collected at the PN junction 17 and detected as an optical signal. Similarly, carriers generated in the right photodiode are collected at the PN junction 18 and detected as an optical signal. At this time, since the two photodiodes are completely separated by the isolation region 14 made of an insulating layer, carriers generated in the left photodiode, for example, do not mix into the right PN junction 18.

Furthermore, carriers generated deeper than the separation region are collected on the reverse biased back electrode 19, and therefore do not diffuse into the N layer 12.

Therefore, optical and physical crosstalk is significantly reduced.

Also, since no current flows between the two photodiodes through the insulating layer, electrical crosstalk is also significantly reduced.

The configuration shown in FIG. 11 is similar to the configuration shown in FIG. 10 except that the substrate is an N ⁻ layer.

The substrate is N ^- manufactured by the pulling method (CZ method).
A silicon substrate 11 having a high resistance of 10 Ωcm is used. Silicon crystals made by the CZ method usually have 5
~50ppm of dissolved oxygen.

An N-type region 12 is formed on this substrate by an epitaxial method.

Then, heat treatment is performed in nitrogen (N ₂ ) gas at 800°C for 1 to 16 hours and in dry oxygen at 1050°C for 18 hours.
This step creates minute defects in the N ^- substrate 11 due to the condensation of oxygen.

Because this microdefect acts as a recombination center,
The lifetime of the carriers generated within the N ⁻ substrate 11 is extremely short, and all of them disappear before they diffuse into the N region 12 . Therefore, all optical crosstalk can be eliminated as in the device shown in FIG.

However, although these methods are effective in eliminating crosstalk, they have problems with dark current and reverse breakdown voltage.

The maximum amount of charge that can be stored in a photodiode is determined by the product of junction capacitance and bias voltage.

Since the junction capacitance exhibits a constant value depending on the type of substrate, the magnitude of the bias voltage determines the maximum amount of charge. When a reverse bias is applied to the P layer on the light-receiving surface, if the bias voltage is too large, the depletion layer that has spread to the N layer may reach the silicon substrate.

At this time, in the configuration shown in FIG. 10, the P layer 17 on the light receiving surface and the P layer 10 on the substrate are short-circuited through the depletion layer, so that current flows from the light receiving surface to the substrate. This causes defects such as a decrease in the amount of photodiode signal charge and poor breakdown voltage. Furthermore, even in the configuration shown in FIG. 11, if the depletion layer spreads in the N ^- region, the N ^-
Defects in the layer lead to deterioration of breakdown voltage and increase in dark current. Furthermore, the depletion layer may extend laterally and reach the isolation region 14.

The isolation region has poor crystallinity due to damage during formation, and when the depletion layer reaches it, leakage current flows intensely. For this reason, the bias voltage is limited, resulting in a limit to the maximum amount of charge detected by the photodiode. Furthermore, it determines the dark current increase accumulation time, causing deterioration of S/N (signal to noise ratio).

In general, the spectral sensitivity characteristics of a photodiode using a silicon semiconductor material as a substrate are such that the photoelectric conversion sensitivity on the long wavelength side is higher than that on the short wavelength side, as shown in FIG.

Further, since the light sources generally used as ultraviolet to infrared light sources in spectroscopic analysis have a large output in the infrared region, the output obtained from the photodiode is also about two orders of magnitude larger in the infrared region than in the ultraviolet region.

Furthermore, infrared light easily becomes stray light, reducing the purity of the signal, so a silicon photodetector with low infrared sensitivity is required.

Since the absorption coefficient of long-wavelength light in silicon is small, it is known that it is absorbed relatively deep in the crystal, creating electron-hole pairs.

In order to effectively remove this problem, see Figures 10 and 11.
It is said to be effective to use the configuration shown in the figure to allow carriers generated deep within the crystal to be absorbed elsewhere or to disappear within a short period of time.

Therefore, the thinner the N layer, which is an effective light absorption layer, is, the more effective this method will be. However, when electrical characteristics other than spectral sensitivity characteristics are considered, the conventional technology has problems in terms of dark current, breakdown voltage, etc.

Furthermore, using a thinner N layer causes an increase in resistance and non-uniformity between the electrodes added to the N layer and each element, resulting in decreased response speed and unevenness in output when used in accumulation mode due to non-uniformity in applied voltage. This is thought to bring about uniformity.

As described above, the conventional device or the conceivable device cannot simultaneously satisfy the prevention of crosstalk and the electrical characteristics, spectral sensitivity characteristics, etc.

(Objective of the Invention) The object of the present invention is to achieve high resolution performance and high position resolution performance by significantly reducing various crosstalks within a photodiode array, and to achieve leakage current when a high bias voltage is applied. The object of the present invention is to provide a semiconductor photodetection device that can reduce the amount of charge and increase the amount of detected charge.

(Structure of the Invention) In order to achieve the above object, a first aspect of the present invention is provided.
A semiconductor photodetection device includes a semiconductor photodetection device in which a plurality of photodiodes each having a junction for photodetection are arranged in an array in the same semiconductor region, a semiconductor substrate having a second conductivity type, and a semiconductor substrate having a second conductivity type; a first semiconductor region having a first conductivity type formed by epitaxial growth on a substrate; and the first semiconductor region provided between the second conductivity type semiconductor substrate and the first semiconductor region. The first semiconductor region has a sufficiently high impurity concentration and is thinner than the semiconductor region of
a plurality of photodiodes of a second conductivity type formed in the first semiconductor region; and a plurality of photodiodes of a second conductivity type formed in the first semiconductor region to isolate the photodiodes from each other. It is configured by providing an isolation region made of a polycrystalline semiconductor having one conductivity type.

A second semiconductor photodetection device according to the present invention is such that the second conductivity type semiconductor substrate is replaced with a first conductivity type semiconductor substrate.

(Example) Hereinafter, the present invention will be described in detail with reference to the drawings and the like.

FIG. 1 is a diagram showing a first embodiment of a semiconductor photodetection device according to the present invention, and FIG. 1A is a cross-sectional view;
Figure B is a plan view.

The semiconductor photodetector according to the present invention has the following features: by forming a separation region 21 made of polycrystalline semiconductor between the junctions 17 and 18 of adjacent photodiodes,
This eliminates crosstalk.

A thin, highly concentrated impurity layer 9 is formed between the substrate 10 and the photodiode region to achieve a reduction in leakage current and an increase in the amount of detected charge when a high bias voltage is applied.

Carriers generated in the photodiode on the left side of FIG. 1 are collected at the PN junction 17 and detected as an optical signal. Similarly, carriers generated in the right photodiode are collected at the PN junction 18 and detected as an optical signal. At this time, since the two photodiodes are completely separated by the isolation region 21 made of an insulating layer, carriers generated in the left photodiode, for example, do not mix into the right PN junction 18.

Since a reverse bias is applied between the P-type substrate 10 and the N-type region 12 during use, carriers generated within the P-type substrate 10 do not diffuse into the N-type region 12 and are all collected on the electrode 19 on the back surface. In this way all optical crosstalk can be eliminated.

Since no current flows between the two photodiodes through the isolation region 21, electrical crosstalk is likewise prevented.

The depletion layer does not spread in the highly concentrated N ⁺ layer 9, and the depletion layer stops spreading when it reaches the N ⁺ layer 9.
The periphery of the isolation region 21 has a high concentration of N ⁺ because the phosphorus-doped polycrystalline silicon acts as a diffusion source.

Even if a bias voltage is applied to the photodiode to the extent that the depletion layer reaches the isolation region 21, the depletion layer does not spread to the highly concentrated N ⁺ layer, and as shown in FIGS.
Leakage current due to damage during formation of the isolation region as in the configuration shown in the figure is not a problem.

In this way, the N ⁺ layer forms a potential barrier due to impurities and also acts as a barrier against the internal electric field. Therefore, the bias voltage applied to the photodiode is not limited, and the thickness of the N layer 12 in the photodiode region can be made as thin as possible.

FIG. 3 shows the spectral sensitivity characteristics of the device.

As is clear from this graph, the difference in output between short and long wavelengths is smaller than that of the conventional device (see FIG. 12).

This makes it easy to connect to external circuits,
The amount of detected charge can also be increased.

Furthermore, since the separation region is formed by plasma etching, the width can be reduced to 2 μm or less. Therefore, the spacing between photodiodes can be reduced, and the resolution performance and decomposition performance of the photodiode array can be improved.

Furthermore, by inserting the N ⁺ layer 9, uniformity of sensitivity and improvement of response speed can be achieved.

That is, since the N ⁺ layer 21 and the N ⁺ region 9 in the isolation region are short-circuited, the silicon dioxide film on the isolation region 21 is removed and the electrode wiring 20 is removed as shown in FIG. 1A.
By doing this, the voltage applied to each element within the photodiode becomes equal at all positions, making it possible to improve uniformity of sensitivity and response speed.

Next, the manufacturing process of the device of the embodiment will be explained with reference to FIG.

(A) First, on a P-type silicon substrate 10, a highly concentrated N ⁺ layer 9 is grown to a thickness of about 1 μm by epitaxial growth, ion implantation, or diffusion, and an N layer 12 is grown to a thickness of 5 μm by epitaxial growth. Then, the silicon dioxide film 13 is formed to a thickness of about 1 μm by thermal oxidation.
Make it grow.

The impurity of the N ⁺ layer is preferably Sb, which has a small diffusion coefficient.

(B) The silicon dioxide film in the portion that will become the isolation region is removed by photo-etching, and the N-type silicon layer 12 is further removed until it reaches the N ⁺ layer 9 using a plasma etching method.

(C) Polycrystalline silicon 21 doped with, for example, phosphorus or arsenic at a high concentration is deposited over the entire surface by CVD, and unnecessary portions are removed.

(D) Next, the silicon dioxide film on the part that will become the light-receiving surface is removed by photoetching, and P is deposited by a diffusion method.
After forming the mold region 15, a silicon dioxide film is grown to a thickness of about 0.2 μm by thermal oxidation.

(E) Predetermined electrode wiring 2 with metal film such as aluminum
0 and the process ends.

FIG. 4 is a sectional view and a plan view showing an embodiment of a second semiconductor photodetecting device according to the present invention.

The structure shown in FIG. 1 uses a P-type substrate, but this embodiment uses a high-resistance N ^- substrate 11.

In this case, as the substrate 11, the pulling method (CZ method)
A silicon substrate with a high resistance of 10 Ωcm and an oxygen concentration of 16 to 18 ppm was used. On this substrate 11, an N ⁺ type region 9 is formed by epitaxial method.
After forming the N-type region 12, heat treatment is performed for 1 to 16 hours in nitrogen (N ₂ ) gas at 800°C and for 18 hours in dry oxygen at 1050°C.

This step creates micro defects in the N ^- substrate 11 due to condensation. This is a known method called the internal getter method.

Because this microdefect acts as a recombination center,
The lifetime of the carriers generated within the N ⁻ substrate 11 is extremely short, and all of them disappear before they diffuse into the N region 12 . Therefore, all optical crosstalk can be eliminated.

Various modifications can be made to the embodiments described in detail above within the scope of the present invention.

In each example device, the same effect can be expected even if the P type and N type are changed.

(Effects of the Invention) As described above in detail, the semiconductor photodetector according to the present invention has a separation region made of polycrystalline silicon between the photodiodes, so that optical
Various types of crosstalk, such as physical and electrical, can be significantly eliminated.

Therefore, there is no mutual interference between the photodiodes, and as a result, resolution performance and decomposition performance can be improved.

Furthermore, the width of the separation region can be reduced to 2 μm or less, and an improvement in resolution performance can be expected at the same time.

Further, since the semiconductor photodetector according to the present invention has a highly concentrated impurity layer between the substrate and the photodiode region, the depletion layer does not spread to the substrate when a bias voltage is applied, and the bias voltage applied to the photodiode is It can be made as large as possible and the sensitivity of the photodiode can be improved.

Furthermore, since the N layer in the photodiode region can be made as thin as possible, the photoelectric conversion sensitivity of long wavelength light is reduced and connection with external circuitry is facilitated. Furthermore, since the isolation region and the high concentration impurity layer are short-circuited, a voltage can be applied to each photodiode through the isolation region. Therefore, the uniformity of the voltage increases, and as a result, the uniformity of sensitivity and the response speed are improved.

The photodetection device according to the present invention can achieve high position resolution performance and high resolution performance by using it as a single unit for position detection and spectrophotometry. Further, by using the semiconductor photodetection device according to the present invention as an image sensor in combination with a self-scanning circuit, clear images can be obtained.

[Brief explanation of the drawing]

FIG. 1 shows a first diagram of a semiconductor photodetection device according to the present invention.
FIG. 2 is a cross-sectional view and a plan view showing an example of the invention. Second
The figure is a sectional view showing the manufacturing process of the device of the embodiment. FIG. 3 is a graph showing the spectral sensitivity characteristics of the device of the embodiment. FIG. 4 is a sectional view showing another embodiment of the semiconductor photodetecting device according to the present invention. Fifth
The figure is a cross-sectional view showing optical crosstalk in a conventional photodiode array type semiconductor photodetecting device. FIG. 6 is a cross-sectional view showing physical crosstalk in a conventional photodiode array type semiconductor photodetection device. FIG. 7 is a sectional view showing one element of a conventional image sensor. FIG. 8 is an equivalent circuit diagram of an image sensor obtained by changing the conventional image sensor shown in FIG. 7 to a P channel MOSFET structure using an N substrate. FIG. 9 is a cross-sectional view showing a configuration in which isolation regions made of a P ⁺ layer are formed between photodiodes to prevent crosstalk. FIG. 10 is a sectional view of a device in which an isolation region made of an insulating layer is formed between photodiodes to prevent crosstalk. FIG. 11 is a sectional view of another device in which an isolation region made of an insulating layer is formed between photodiodes to prevent crosstalk. FIG. 12 is a graph showing the spectral sensitivity characteristics of a conventional device. 1... N-type silicon substrate, 2... Source region,
3...Drain region, 4...Gate electrode, 9...
N ⁺ layer, 10...P-type silicon substrate, 11... ^N-
Silicon substrate, 12... N type region, 13... Silicon dioxide film, 14... Separation region, 15, 16... Light receiving surface (P region), 17, 18... PN junction, 1
9... Back electrode, 20... Surface electrode, 21... Polycrystalline silicon, 22... Surface electrode.

Claims (1)

[Scope of Claims] 1. A semiconductor photodetection device in which a plurality of photodiodes having junctions for photodetection are arranged in an array in the same semiconductor region, a semiconductor substrate having a second conductivity type, and a semiconductor substrate having a second conductivity type; a first semiconductor region having a first conductivity type formed by epitaxial growth on a substrate; and the first semiconductor region provided between the second conductivity type semiconductor substrate and the first semiconductor region. a thin layer of a first conductivity type having a sufficiently high impurity concentration and thinner than the semiconductor region of the semiconductor region; a plurality of photodiodes of a second conductivity type formed in the first semiconductor region; 1. A semiconductor photodetecting device, comprising: an isolation region made of a polycrystalline semiconductor having a first conductivity type formed within the first semiconductor region so as to isolate them from each other. 2. In a semiconductor photodetection device in which a plurality of photodiodes having junctions for photodetection are arranged in an array in the same semiconductor region, a semiconductor substrate having a first conductivity type and an epitaxial growth layer on the semiconductor substrate are provided. a first semiconductor region having a first conductivity type formed by the substrate, and a sufficiently higher impurity concentration and thickness than the first semiconductor region provided between the substrate and the first semiconductor region. a thin layer of a first conductivity type formed in the first semiconductor region; a plurality of photodiodes of a second conductivity type formed in the first semiconductor region; The first formed
What is claimed is: 1. A semiconductor photodetection device comprising an isolation region made of a polycrystalline semiconductor having a conductivity type. 3. The semiconductor photodetecting device according to claim 2, wherein the substrate is manufactured by a CZ method.