JPS61231772A - Manufacture of semiconductor device - Google Patents

Abstract

PURPOSE:To prevent deterioration caused by light application and make the width of the depletion layer of the activation region of an I-type semiconductor as wide as 1mum or wider along the direction of the current flow of a photoelectric conversion device with a PIN junction by a method wherein a non-single crystal semiconductor, including an amorphous semiconductor, containing hydrogen or halogen is annealed by light and column shape grains with anisotropy are formed by promoting crystallization. CONSTITUTION:A light permeable conductive film, which is the first conductive film 2, is formed on the whole surface of a substrate with an insulating surface, for instance a glass substrate 1. Then a laser beam is applied to form the first open ditches 13 for scribe lines and the first electrodes 15 are formed in respective activation element regions 31 and 11. After that, a non-single crystal semiconductor 3 is formed over the whole surface. Further, the second open ditch 14 is formed over the left side of the first open ditch 13. High-intensity light with a wavelength of 500nm or longer is applied to the non-single crystal semiconductor to form an I-type semiconductor as a semiconductor with column shape grains whose grain direction is vertical to the substrate. At that time, the semiconductors of the passivation regions 34 and 33' in the ditch 13 are kept from annealing by a metal mask 50 to provide isolation between the first electrode of the element 31 and the first electrode of the element 11.

Description

Detailed Description of the Invention "Field of Application of the Invention"
Relating to a photoelectric conversion device that generates a high voltage by electrically connecting in series a plurality of photoelectric conversion elements (simply referred to as elements) in which a non-single crystal semiconductor including an amorphous semiconductor having an N junction is provided on a substrate having an insulating surface. .

"Prior Art" Conventionally, amorphous semiconductors are known as non-single crystal semiconductors to which hydrogen or halogen elements are added. However, since such a semiconductor has an amorphous structure and does not actively use crystallinity, the thickness of the carrier depletion layer of the ■-type semiconductor layer in the PIN junction is as narrow as 0.3 μm or less, and Deterioration occurred due to light irradiation at 100 m (7 cm2).

"Problems to be Solved by the Present Invention" The present invention aims to optically anneal a non-single crystal semiconductor containing hydrogen or a halogen element, including such an amorphous semiconductor, to crystallize columnar grains having anisotropy. In order to prevent deterioration due to light irradiation, the depletion layer in the active region of the ■-type semiconductor is made thicker (wider) by 1 μm or more in the direction of current flow in a photoelectric conversion device having a PIN junction. The purpose is to

Furthermore, the non-active semiconductor region constituting the connecting portion is made to have high resistance to prevent leakage between electrodes in this non-active region.

"Means for Solving the Problem" The present invention provides a means for solving the problem by transmitting light to a non-single-crystal semiconductor directly or through the transparent electrode from the transparent electrode side.
By irradiating pulsed strong light with a wavelength of 500 nm or more with a light-shielding metal mask in between, hydrogen or halogen elements are applied to the I-type semiconductor layer or the P- or N-type semiconductor layer adjacent to it at the grain boundaries of columnar grains. It segregates a large amount and stores it internally while promoting crystallinity.

In particular, the present invention is suitable for use with a wavelength of 500 nm or more, where the light absorption is small.
YAG below μm, e.g. 0.53μ or 1.06μ
Irradiate pulsed strong light from a laser, a ruby laser with a wavelength of 0.69μ, or a xenon lamp to promote the crystallinity of the ■-type semiconductor as a whole or to a sufficiently deep region in the form of columnar grains with anisotropy. , so-called optical annealing was performed. For this reason, the light can perform optical annealing to a deep region where the optical absorption coefficient of the semiconductor is relatively small.
A wavelength of nm or more was used.

In the present invention, the increase in electrical conductivity that accompanies this optical annealing must not impede isolation in an integrated structure. Therefore, in the method of the present invention,
This photoannealing was performed only in the active semiconductor region, so that the grain boundaries were parallel to the current flow and the inactive region had a layer with substantially higher resistance than the active region.

In the present invention, a YAG laser is applied to a non-single crystal semiconductor with laser light (Q-switch) in order to form a connection part for integration in an inactive region in a process that is performed simultaneously with, before, or after annealing. It was scribed and removed using light.

"Function" Since the non-active region in the semiconductor of the photoelectric conversion device produced according to the present invention is not photo-annealed, it has a substantially higher resistance than the active region that has been photo-annealed, and has sufficient iso-isolation. You can do rations. On the other hand, the current in PIN is in the direction along the columnar grains formed by photoannealing, and the current does not cross grain boundaries. In addition, an optical confinement type cell is formed in which light is randomly given optical paths within one layer and crosses grain boundaries containing a large amount of hydrogen, which has a large optical absorption coefficient, many times. That is, it is effective to make the first and second electrodes on the back side uneven and to make the back electrode reflective. The present invention has these effects, and it is said that the manufacturing of an integrated photoelectric conversion device can be completed only by a photo-annealing process in which light is applied from the non-single-crystal semiconductor surface side, without any other extra steps. It has characteristics.

The arrangement, size, and shape of elements in the device of the present invention are determined by design specifications. However, in order to simplify the content of the present invention, in the following detailed description, the first electrode on the lower side (substrate side) of the first element and the second electrode of the second element disposed on the right side thereof will be described. The pattern is based on the case where the electrodes (on the semiconductor, that is, on the side away from the substrate) are electrically connected in series.

Then, a laser beam for LS, for example, a YAG laser (focal length 40 mm) having a wavelength of 1.06 μ (light diameter approximately 50 μ) or 0.53 μ (light diameter approximately 25 μ) is irradiated to this defined position.

Further, it is moved at an operating speed of 0.05 to 5 m/min, for example, 30 cm 2 /min, to create a dependent relationship with the previous process.

In the present invention, when the substrate is a light-transmitting glass, a non-single-crystal semiconductor is formed on the non-light-transmitting substrate, and the surface of the non-single-crystal semiconductor is optically annealed with light of 500 nm or more and 5 μm or less (the energy density is lXl0' to IX10'W/cm2, and the energy density during laser scribing is 5
~5 1/10~1/103 from X10'W/cm2
This is a breakthrough in that by simply adding a photo-annealing process to the manufacturing process of conventionally known photoelectric conversion devices, the yield can be increased from the conventional approximately 60% to 84%. An object of the present invention is to provide a method for manufacturing a photoelectric conversion device.

The details of the invention are shown below in accordance with the drawings.

"Example 1" FIG. 1 is a longitudinal cross-sectional view of a photoelectric conversion element produced according to the present invention.

In the drawings, a substrate having an insulating surface, such as a glass substrate (1), having a diameter of 1.5 cm and a width of 2 cm was used.

Further, on this upper surface, a first conductive film (2) is formed over the entire surface.
A transparent conductive film having a thickness of 0.1 to 0.5 μm was formed. This conductive film had unevenness with a height difference of 0.1 μm.

This transparent conductive film (2) is a transparent conductive film whose main component is tin oxide to which a halogen element such as fluorine is added, or ITO (tin oxide indium) (500 to 5000
(typically 500 to 1,500 people) was formed by sputtering or spraying to form the first conductive film.

Thereafter, a non-single-crystal semiconductor layer to which hydrogen or a halogen element is added, which has a PIN junction, that is, a non-single-crystal semiconductor that generates photovoltaic force upon irradiation with light, is formed on the upper surface by a plasma CVD method, a photo-CVD method, or an LPCVD method. (3)
The minimum oxygen concentration in the I-type semiconductor is 5×IQ18cm-3
The thickness was 0.3 to 5.0 μm, typically 2.0 μm.

For example, a P-type (stXcl-X O<x<1) semiconductor (
(approximately 200 persons) - Type 1 amorphous or semi-amorphous silicon semiconductor (approximately 2.0 μm) - Non-single crystal semiconductor (3) with one PIN junction consisting of a semiconductor having N-type microcrystals (approximately 500 persons) The film was formed to have a uniform thickness.

In FIG. 1, a second conductive film (5) is further formed on this upper surface.
and a connector (30) was formed.

The structure to be irradiated is the one shown in FIG. 1 or the substrate on which the semiconductor (3) is laminated before forming the back electrode.

The irradiation area of the light source was set to 1 mm x 9 mm using an optical system using a ruby pulse laser beam.

In addition, in relation to the scan speed, the thickness is set to 1 mm, and in order to control the irradiation energy density, the thickness is 100 μm ~
It may be varied up to 3III+.

Here, a laser oscillator manufactured by CTC Comtic Training Co., Ltd. was used.

Furthermore, this laser beam is focused into a rectangular shape by a lens, has pulsed light (frequency 300Hz to 30KHz), and has a power output of 5KW/
An! (In case of width 1m1I+).

This irradiation light (25) was irradiated onto the irradiated surface of a moving base at a constant speed. At this time, a metal mask was used to prevent the previously inactive region from being irradiated with light for photoannealing.

In this way, the total thickness of one layer in a non-single crystal semiconductor has a wavelength of 0.
．． When using a wavelength of 7 μm) or 0.53 μm, it was possible to create a region that promotes crystallization of columnar grains in a region with a depth of 3,000 to 5,000 people, which is about half that. The fact of this crystallization was found by performing laser Raman spectroscopy after this step. In addition, since the annealing method of the present invention is a light pulse annealing, hydrogen or halogen elements that are already contained are less likely to be degassed to the outside during crystallization, and they are segregated at grain boundaries and are It can neutralize the unpaired bond in the field. In addition, since crystallinity or orderliness is promoted by photoannealing, photodegradation characteristics are reduced, and in addition, the width of the depletion layer in rH in an amorphous semiconductor between PNs is
In order to have crystallinity from 0.3 μm in the bond,
It had the double feature of being able to be stretched to 1 to 3 μm. Therefore, the optimum thickness of one layer can be increased from 0.5 μm of an amorphous semiconductor to 1.5 to 2.0 μm, and the current as a photoelectric conversion device can be increased.

In this way, the second conductive film (5) on the semiconductor shown in FIG.
) used metal and a transparent conductive oxide film (CTF). The thickness of each layer was formed by 300 to 1,500 people, and the structure was an optical confinement type.

As this CTF, a light-transmitting conductive film made of a non-oxide conductive film such as a chromium-silicon compound may be used.

These are electron beam evaporation method or sputtering method,
A CVD method including a VD method and a photo-plasma CVD method was used to form the semiconductor layer at a temperature of 250'C or less in order not to deteriorate the semiconductor layer.

Thus, for the irradiated light (1o), on the substrate (3.5mm x 3cm) as in this example, 11.8% (1o)
, 05 cm2).

FIG. 2 is a configuration diagram illustrating the outline of the idea of the present invention.

Figure 2 (A) shows columnar grains (40) in a ■-type non-single crystal semiconductor.
(only three locations are shown here) whose column diameter (41) is 0
．． When examined using SEM (scanning electron micrograph) at 1 to 2 μm, it was found that the growth was vertical.

Furthermore, the crystal grain size is sequentially larger on the side irradiated with strong light than on the opposite side.

This energy band width is shown in Figure 2 (B), (C),
(D), shown in half. In other words, Fig. 2 (B) is similar to Fig. 2 (
This is the case where no electric field is applied in the vertical direction of the central part of the columnar grains of B-B' in A). In this way, since the amount of hydrogen desorbed is larger on the side irradiated with strong light than on the opposite side, the energy band width becomes narrower, and light is irradiated from the transparent insulating substrate (1) side as a photoelectric conversion device. In this case, even in type I non-single crystal semiconductors, the energy band width is essentially wide-to-narrow, and the electrons and holes generated by light irradiation can be extracted more efficiently (extracted as electric power to the outside). (C) is a hole (42) and an electron (4) due to the electric field.
3) is drifting. (D) shows the energy band in the cross section of columnar grains taken along line AA' in the lateral direction. Then, electrons (42') and holes (4
3') generates carriers (42) and (42'') through photo-electrical conversion in the region where a large amount of hydrogen exists at the grain boundary.

The carriers drift to a more stable level inside the columnar grains.

Therefore, carriers can be drifted in the center of the columnar grains, which are separated by the grain boundaries where most of the recombination centers are present.

On the other hand, since photoannealing is not performed in the isolation region of the connection part in the integrated structure, the resistance is higher than in the region where grain boundaries are formed by photoannealing, and it is difficult for leakage current to flow therein. That is, when attempting to integrate a photoelectric conversion device, high annealing should not be performed in order to prevent leakage current from flowing in the inactive region forming the connection portion.

In FIG. 2, hydrogen segregates at grain boundaries (45) of columnar grains (40) and becomes highly concentrated. Therefore, this grain boundary increases the optical Eg and has a large light absorption coefficient like an amorphous semiconductor.

This shows that it is important for light to cross this grain boundary many times in order to efficiently perform light-to-electrical conversion with a small thickness.

Furthermore, for these reasons, it is effective to set the diameter of the pillars of the semiconductor to be smaller than the thickness of the intrinsic or substantially intrinsic semiconductor, which is 0.1 to 5 μm, and generally to 0.05 to 0.2 μm. It is.

If it is more than this, the carriers generated at the grain boundaries will drift to the P or N layer before drifting into the interior of the columnar grains, and this will not be carried out sufficiently, meaning that the meaning of raising the barrier with hydrogen at the grain boundaries will be reduced. Resulting in.

(15) and (16) in FIG. 3 indicate the vertical electrical conductivity and the lateral electrical conductivity of the I-type non-single crystal semiconductor, respectively.

The initial values in each graph are the photoconductivity (15) investigated in the structure of an amorphous semiconductor doped with hydrogen,
(17) and dark conductivity (16), (18). The next plot shows the characteristics after performing ruby laser annealing on the amorphous semiconductor to polycrystallize it, and the third plot shows the electrical conductivity after irradiation with light irradiation (AMI") for 2 hours. Furthermore, the fourth measurement point shows the characteristics after thermal annealing at 150° C. for 2 hours.

It is clear from these results that the electrical conductivity in the vertical direction improves approximately three times by photoannealing;
The electrical conductivity shown in (17) in the horizontal direction across grain boundaries decreases to about 1710.

In addition, photodeterioration (Stebler-Lonski effect) is observed in the electrical conductivity characteristics in the lateral direction due to hydrogen, oxygen, etc. present at the grain boundaries, but there is virtually no change in the electrical conductivity in the vertical direction. , is extremely stable.

In addition, performing optical annealing from the second electrode side after forming a transparent conductive film (for example, ITO) as the electrode of the cylinder 2 after forming a non-single crystal semiconductor can reduce desorption of hydrogen, halogen elements, etc. due to optical annealing. Furthermore, it was possible to widen the light intensity and the light irradiation time, which was effective.

Example 2 FIG. 4 is a longitudinal sectional view showing the manufacturing process of the present invention.

The manufacturing process of the film was carried out in the same manner as in Example 1.

That is, in the drawing, a substrate having an insulating surface, such as a glass substrate (1), whose length is in the left-right direction in the drawing) I
Dcm and a width of 10 cm were used. Further, on this upper surface, a light-transmitting conductive film (2) which is a first conductive film (2) is applied over the entire surface.
2) was formed with unevenness to a thickness of 0.1 to 0.5 μm.

After that, a YAG laser (wavelength 0,
A 53 μm (pulse width: 30 ns) processing machine (Nihon Denki type) applies an average output of O03~IW (focal length: 40 mm), focuses a laser beam with a diameter of 5IIllllφ, and produces a spot diameter of 20~70 μmφ, typically 50 μmφ. Controlled by a microcomputer, a laser beam is irradiated from above and scanned to form a first opening (13) for a scribe line, and each active element area (31),
Laser scribe (L) the first electrode (15) on (11).
(referred to as S).

The openings (13) formed by LS had a width of about 50 μm and a length of 10 cm, and the depths were completely cut and separated to form the first electrodes.

Thus the first element (31) and the second element (11)
The width of the area constituting the area is 5 to 40 mm, for example 10 mm.

At this time, the semiconductor filled in the first semiconductor (13) is of a high resistance type, and isolation must be performed.

Thereafter, a non-single-crystal semiconductor layer (3) is deposited on the upper surface to a thickness of 0.5 to 5.0 μm, typically 2.0 μm, by plasma CVD, photoCVD, or LPCVD, as in the example.
It was formed to a thickness of μm.

Furthermore, as shown in FIG. 4(B), the first open groove (1
3) across the left direction side (first element side).
(14) was formed by the second LSI process.

In this drawing, the first and second open grooves (13), (1
4) are shifted by 100μ between the centers.

The second open groove (18) thus exposed the side or top surface (8), (9) of the first electrode.

Furthermore, the ruby pulse light laser annealing which emits light with a wavelength of 500 nm or more in the method of the present invention was performed in the same manner as in Example 1.

The substrate to be irradiated has a structure as shown in FIG. 4(B).

Prevented the running light from being irradiated.

Although not shown in the drawings, the non-single crystal semiconductor was irradiated with intense light to form an I-type semiconductor in the semiconductor having columnar grains perpendicular to the substrate.

At this time, the inactive area (34) (33'
) is no longer performed.

In FIG. 4, as shown in FIG. 4(C), a second conductive film (5) and a connector (30
) was formed.

This second conductive film (5) consists of a metal and a transparent conductive oxide film (C
TF) was used. Its thickness is 300~150 respectively
0 people formed it.

In addition, in the state shown in FIG. 4(C), when a transparent conductive film (for example, ITO) is used as the second conductive film, photoannealing is performed on the non-conductive material sandwiched between the first and second electrodes. This was an effective method because it reduced the amount of hydrogen or halogen elements desorbed from single crystal semiconductors.

Next, a third LS was used to cut and separate the plurality of second electrodes (39) and (38) to form a third electrode (20) and isolate them.

As this CTF, a light-transmitting conductive film made of a non-oxide conductive film such as a chromium-silicon compound may be used.

Thus, as shown in FIG. 4(C), a plurality of elements (
31) and (11) were connected in series at the connecting part (4).

FIG. 4(D) shows the present invention further completed as a photoelectric conversion device. That is, as a passivation film, a silicon nitride film (21) is uniformly formed to a thickness of 500 to 2,000 layers by a plasma vapor phase method or a photo plasma vapor phase method, and the leakage current between each element is caused by adsorption of moisture, etc. This further prevented the outbreak.

Furthermore, external lead-out terminals (22) and (22') were provided at the periphery.

Thus, for the irradiation light (10), on a substrate (locm x 10 cm) as in this example, each element is
It is provided in a strip shape of 0 mm x 92 mm, and the width of the 200 μm external extraction part in the connecting part is 3 mm, and the peripheral part is 4 m.
m, there were essentially 10 stages within 88 mm x 92 mm.

As a result, the panel was 8.3% (theoretically it would be 9.6%, but the effective conversion efficiency was reduced due to the 11-stage series-connected resistance. It was possible to have an output power of .7W.

Furthermore, make this panel larger, for example, 40cmX.
A 120cm x 40cm NE is packaged by combining two 60cm pieces in series within a rigid frame made of aluminum sash or a flexible frame made of carbon blank.
[It is possible to provide a high power panel of IQ standard.

In addition, for this NEDO standard panel, the top surface of the photoelectric conversion device of the present invention is attached to the back surface of the glass substrate (opposite side to the irradiation surface) using Seaflex to increase mechanical strength against wind pressure, rain, etc. is also valid.

The effective efficiency of the panel is AMI (100mW/cm
2), we were able to obtain 8.7%, output cuff, and 8W.

The effective area is 82.8 cm”, and the total panel area is 82.8 cm”.
8% could be used effectively.

For laser annealing in the present invention, a YAG laser with a 0.53 μm pulse width of 30 ns or a ruby laser with a wavelength of 0.69 μm (pulse width 70 ns) was used.

However, this 50Or+m (0.5μm) ~ 5000nm
It is effective to use another laser beam or a flash-shaped xenon lamp to emit light with a wavelength of (5 μm).

In addition, it goes without saying that the present invention may be applied to photosensors and image sensors to which hydrogen or halogen elements are added and have a PIN or NIP structure.

Further, the strong light for optical annealing may be irradiated on either side of the substrate.

"Effect" As shown in FIG. 5, the semiconductor layer formed by the method according to the present invention is
) is extremely stable. As an example, the deterioration characteristics of a general amorphous PIN type semiconductor (50)
In comparison, in the case of the type 2 semiconductor of the semiconductor device formed according to the present invention, its deterioration is extremely small despite its thickness of 1.0 to 5 μm. (51) is Example 1, (52)
is the characteristic of Example 2.

Furthermore, the first non-active region of the photoelectric conversion tA device formed according to the present invention, that is, the connection part in the integrated structure, is a non-single crystal semiconductor that retains the initial structure, for example, an amorphous semiconductor having a high resistance type, and is a photoactive The region becomes a semiconductor with a high degree of crystallinity by photo-annealing, and the difference in film quality makes it difficult for leakage current to flow in the lateral direction of the connecting portion.

Furthermore, performing high-temperature annealing after forming the second electrode resulted in less hydrogen desorption and was more effective.

[Brief explanation of drawings]

FIG. 1 shows a longitudinal cross-sectional view of a photoelectric conversion device produced according to the present invention. FIG. 2 shows the columnar grains obtained according to the invention and the corresponding energy band diagram. FIG. 3 shows the characteristics obtained with the present invention. FIG. 4 is a longitudinal sectional view showing the manufacturing process of a photoelectric conversion device created according to the present invention. FIG. 5 shows the light irradiation reliability characteristics of a photoelectric conversion device not subjected to optical pulse annealing according to the present invention and a photoelectric conversion device subjected to optical pulse annealing according to the present invention.

Claims (2)

[Claims] 1. A substrate having an insulating surface has a first electrode, a non-single crystal semiconductor doped with hydrogen or a halogen element having a PIN junction in close contact with the electrode, and a second electrode on the semiconductor. In a semiconductor device in which a plurality of photoelectric conversion elements are connected in series at a connecting portion, strong light having a wavelength of 500 nm to 5 μm is applied to the non-single crystal semiconductor, and a light shielding mask is provided between the substrate and the strong light. 1. A method for producing a semiconductor device, characterized by irradiating only a photoactive region. 2. 2. A method for manufacturing a semiconductor device according to claim 1, wherein the inactive region constituting the connecting portion is maintained in a structure in which intense light annealing is not performed and is left as a high resistance region.