JP2016092243A - Photoelectric conversion element - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a photoelectric conversion element suitable for irradiation of a monochromatic light with high efficiency.SOLUTION: A photoelectric conversion element 10 comprises: a light absorption member 12; a band-pass filter 14 that is formed to a front surface of a light-receiving surface side of the light absorption member 12; a surface electrode 16 that is formed to a circumference of the band-pass filter 14; a diffusion reflection surface 12b that is the front surface of a back surface side of the light absorption member 12, and is formed at a position opposite to the band-pass filter 14 and the surface electrode 16; and a back surface electrode 18 that is formed so as to cover a whole or a part of the front surface of the diffusion reflection surface 12b. The band-pass filter 14 has a function for selectively transmitting a monochromatic light of at least wavelength λ. The diffusion reflection surface 12b has a function for diffusing an incident light in a parallel direction to the light-receiving surface by a multiple reflection. Each of the surface electrode 16 and the back surface electrode 18 has a function as a collector extracting a carrier, and a function as a reflection coating for reflecting the incident light.SELECTED DRAWING: Figure 1

Description

The present invention relates to a photoelectric conversion element, and more specifically, the energy of monochromatic light (for example, monochromatic light having a high power density of 10 W / cm ² or more, particularly ˜1 kW / cm ² or more from a solar light excitation laser) is converted into electric energy. The present invention relates to a photoelectric conversion element that is suitable for conversion into a light source.

  Non-Patent Document 1 describes an attempt to convert light having a wavelength of 808 nm from a semiconductor laser into electric energy using a gallium arsenide solar cell that has been conventionally used as a solar cell for space use. However, it is pointed out that the conversion efficiency is insufficient due to the combination of the ready-made elements.

On the other hand, efforts are also being made to increase conversion efficiency in conventional solar cells. Specifically, the following methods are known as methods for improving the conversion efficiency of a conventional solar cell.
(A) A method of increasing the conversion efficiency using an optical confinement technique.
(B) A method of suppressing carrier recombination.
(C) A method for avoiding the formation of a Schottky junction at the interface between the semiconductor substrate and the electrode.

(A) Optical confinement technology:
In a conventional solar cell, an uneven structure is provided on the light receiving surface of a semiconductor substrate in order to suppress light reflection on the light receiving surface. When the concavo-convex structure is provided on the light receiving surface, the light reflected by the convex portions of the light receiving surface is reabsorbed by the adjacent convex portions. Various proposals have been made for the concavo-convex structure provided on such a light-receiving surface and the manufacturing method thereof.

For example, Non-Patent Document 2 describes a non-reflective surface shape composed of 2500 regular inverted pyramids per 1 mm ² as an optical confinement technique. Patent Document 1 describes that a non-reflective surface shape is produced by a photoetching method.
Patent Document 2 discloses an alkali etching solution containing monosulfonic acid or a salt thereof, an alkali compound, and water as an etching solution that can form a uniform non-reflective surface shape (pyramid structure) without using a photomask. It is disclosed.

In Patent Document 3, as a method of forming an uneven structure,
(A) A damage layer is formed on the surface of the semiconductor substrate by a mechanical damage formation method such as wire slicing, sandblasting, sandpaper polishing, or chemical damage formation method such as impurity diffusion, ion bombardment, radical irradiation,
(B) Next, a method for performing acid etching on the surface of a semiconductor substrate is disclosed.
In Patent Document 4, as a method of forming an uneven structure,
(A) A groove is formed by irradiating the surface of the silicon substrate with laser light,
(B) A method of forming irregularities by chemical etching with an acid contained in a ratio of hydrofluoric acid: nitric acid = 3 to 10: 1 is disclosed.

(B) Suppression of carrier recombination:
In many conventional solar cells, for example, as disclosed in Patent Document 1, a first diffusion layer (emitter layer) having a conductivity type different from that of the semiconductor substrate is provided on the light receiving surface side of the semiconductor substrate. Is formed. When the emitter layer is formed on the surface of the semiconductor substrate, a pn junction surface is formed at the interface between the two. The pn junction surface serves as a driving force for separating and accelerating the carriers generated by light in the respective directions of the two electrodes (+ electrode and −electrode) of the solar cell. Since the density of carriers generated by light is large near the light-receiving surface, carriers can be efficiently extracted from the two electrodes by providing the emitter layer on the light-receiving surface side.

  In a solar cell in which a pn junction surface is provided near the light receiving surface, a second diffusion layer (back surface field (BSF) layer) is provided on the surface (back surface) opposite to the light receiving surface of the semiconductor substrate. Often provided. The BSF layer refers to a layer doped with a high concentration of impurities having the same conductivity type as that of the semiconductor substrate. The BSF layer is a minority carrier generated when the semiconductor substrate absorbs light, and is originally diffused in the direction of the pn junction surface provided in the vicinity of the light receiving surface, but diffused in the direction of the electrode provided on the back surface. It works to suppress movement (return to the semiconductor substrate side).

  On the other hand, Patent Document 5 discloses a solar cell in which both a first diffusion layer and a second diffusion layer are formed on the back side of a semiconductor substrate. A metal electrode is generally used as the electrode of the solar cell, but the metal electrode reflects light to the outside of the solar cell. By forming both the first diffusion layer and the second diffusion layer on the back surface, it is not necessary to provide an electrode on the light receiving surface side, so that the decrease in efficiency due to the electrode reflecting light is suppressed to some extent. Can do.

Such first and second diffusion layers are generally formed by an impurity doping method. Various methods are known as a method of forming the diffusion layer.
For example, Non-Patent Document 3 includes
(A) A method of applying a coating solution containing impurities to a semiconductor substrate and doping by heat treatment,
(B) A method of inserting a semiconductor substrate into an electric furnace or the like, feeding a gas containing impurities into the electric furnace, and diffusing impurity atoms generated by the decomposition of the gas in the electric furnace (gas phase Diffusion method)
Is disclosed.
For vapor phase diffusion,
(A) a method using a gas having an impurity atom literally in its molecule as an impurity source;
(B) Although the impurity source is a liquid, the carrier gas such as nitrogen is passed through the impurity source so that the impurity source gas is included in the carrier gas and supplied to the surface of the semiconductor substrate.
and so on.

For example, when silicon is doped with phosphorus, which is an n-type impurity,
(A) A method in which a liquid organic compound containing phosphorus is applied onto a wafer with a spinner and heat-treated at 800 ° C. (Patent Document 6),
(B) A method in which a coating solution containing diphosphorus pentoxide (P ₂ O ₅ ) is applied on a wafer and heat-treated at 800 to 1200 ° C. (Patent Document 7),
Etc. are known.

  Further, the ion implantation method is a method in which impurities are accelerated in the form of ions and implanted into a semiconductor substrate. The ion implantation method is generally used as a method of forming a shallow impurity layer with precise control of impurity concentration and depth in a large-scale integrated circuit (LSI) or the like, but manufacturing a photoelectric conversion element (solar cell). The example used for this is also known (Patent Document 8).

(C) Avoidance of Schottky junction formation:
When a Schottky junction is formed at the electrode / semiconductor substrate interface, conversion efficiency decreases. Therefore, in the photoelectric conversion element, it is necessary to avoid the formation of a Schottky junction.
In order to avoid the formation of the Schottky junction, generally, a method of selecting an optimum electrode material according to the composition of the semiconductor substrate is employed. For example, when the semiconductor substrate is silicon and the second diffusion layer is formed on the back surface side, aluminum is used as an electrode material connected to the second diffusion layer (Non-Patent Document 4).

As described above, in the conventional solar cell, various methods for improving the conversion efficiency have been proposed. However, high conversion efficiency cannot be obtained simply by diverting these techniques to a monochromatic solar cell.
Furthermore, a photoelectric conversion element using monochromatic light as incident light and having a concavo-convex structure (diffuse reflection surface) on the surface opposite to the light receiving surface of the light absorbing member, and ideal for such a photoelectric conversion element There has never been proposed an example of a concavo-convex structure and a manufacturing method thereof.

JP 2000-022185 A JP 2013-236027 A Japanese Patent Laying-Open No. 2005-340643 JP 2003-258285 A JP 2013-183004 A JP 2013-012532 A JP 2011-171600 A JP 2013-197538 A

Nobuki Kawashima, “Wireless Energy Transmission Technology-4, Laser Transmission Technology”, The Institute of Electrical Engineers of Japan, Vol. 129, No. 7, pages 422-425 (2009) Kunio Uemura, Tatsuo Saga, Yasunobu Matsutani, “Single-crystal silicon solar cell for space use”, Sharp Technical Report, No. 70, pp. 59-64 (2009) S. M.M. G, “Semiconductor Devices 2nd Edition, Basic Theory and Process Technology” (Industry Books, 2004), Chapter 13 Impurity doping, pp. 402-434 Edited by the Ceramic Society of Japan, “Solar Cell Materials” (Nikkan Kogyo Shimbun, 2006)

An object of the present invention is to provide a highly efficient photoelectric conversion element suitable for monochromatic light irradiation.
Another problem to be solved by the present invention is a high conversion efficiency in a photoelectric conversion element using monochromatic light as incident light and having a concavo-convex structure on the surface opposite to the light receiving surface of the light absorbing member. It is an object of the present invention to provide a concavo-convex structure suitable for obtaining the above.
Furthermore, another problem to be solved by the present invention is to suppress the carrier recombination and avoid the formation of a Schottky bond at the light absorbing member / electrode interface in such a photoelectric conversion element. .

In order to solve the above problems, a photoelectric conversion element according to the present invention has the following configuration.
(1) The photoelectric conversion element includes a light-absorbing member made of a light-absorbing material capable of absorbing monochromatic light having a wavelength λ ₀ and generating carriers, and substantially parallel rays made of the monochromatic light as incident light. Is used.
(2) The photoelectric conversion element is
The light absorbing member;
A band pass filter formed on the light receiving surface side surface of the light absorbing member;
A surface electrode formed around the bandpass filter;
A surface on the back side of the light absorbing member, a diffuse reflection surface formed at a position facing the band pass filter and the surface electrode;
A back electrode formed so as to cover all or part of the surface of the diffuse reflection surface.
(3) The band-pass filter has a function of selectively transmitting at least light having the wavelength λ ₀ .
(4) The diffuse reflection surface has a function of diffusing the incident light in a direction parallel to the light receiving surface by multiple reflection.
(5) The front electrode and the back electrode each have a function as a current collector for extracting the carriers and a function as a reflective film for reflecting the incident light.

The photoelectric conversion element according to the present invention preferably has the following configuration.
(6) The thickness of the light diffusion part of the light absorbing member is 1/200 or more and 1/10 or less of the penetration depth of the incident light into the light absorbing material.
Here, the “light diffusing portion” means that when the light absorbing member is viewed from the normal direction of the light receiving surface, the projection surface of the bandpass filter and the surface electrode and the projection surface of the diffuse reflection surface are An overlapping part.
The “depth of penetration of incident light into the light absorbing material” is a depth at which the intensity of the incident light becomes 1 / e (e is the number of Napier) when the incident light is incident on the light absorbing material. Say.
(7) The shortest distance from the edge of the bandpass filter to the edge of the light diffusing unit is at least twice the diffusion distance.
Here, the “diffusion distance” means that the band-pass filter and the diffusive reflection surface are formed on the front and back surfaces of the light-absorbing material having a thickness corresponding to the thickness of the light diffusion portion, respectively. When the incident light is diffused in a direction parallel to the light receiving surface while multiple reflection of the incident light is performed between a band pass filter and the diffuse reflection surface, the intensity of the incident light is 1 / e (e is , Napier number).

  The diffuse reflection surface is preferably one in which the reflection spectrum of the incident light satisfies the relationship of formula (1) described later.

The photoelectric conversion element according to the present invention preferably has the following configuration.
(8) At least one of the front electrode and the back electrode is
A first layer made of a first conductive material formed on a surface of the light absorbing member and capable of forming an ohmic junction with the light absorbing member;
A laminated structure with a second layer made of a second conductive material formed on the surface of the first layer and having a higher reflectance of the incident light than the first conductive material;
The thickness of the first layer is not less than three times the penetration depth of carriers into the first conductive material, and is not more than the penetration depth of the incident light into the first conductive material.
Here, “the penetration depth of the carrier into the first conductive material” is assumed that the carrier (A) in the light absorbing material and the carrier (B) in the first conductive material can be distinguished, When the concentration distribution in the first conductive material of the carrier (A) in the light absorbing material is calculated using Poisson's equation, the concentration of the carrier (A) in the first conductive material is The depth which becomes 1 / e (e is the Napier number) of the concentration of the carrier (A) in the absorbent material.
“Invasion depth of incident light into the first conductive material” means that when the incident light is incident on the first conductive material, the intensity of the incident light is 1 / e (e is the number of Napier). Say depth.

  A bandpass filter has been conventionally used to extract light of a specific wavelength from white light. When this band pass filter is provided on the light receiving surface of the photoelectric conversion element and monochromatic light is incident on the light receiving surface, the band pass filter again reflects the reflected light from the back surface inside the light absorbing member without blocking the incident light. Reflect on. Thereby, even if the light absorbing member is thinned, a high light absorption rate is secured. Further, the material can be reduced and the tact time can be shortened, so that the cost can be reduced.

  In addition, if the periphery of the bandpass filter is covered with a surface electrode of a predetermined size and a diffuse reflection surface and a back electrode are provided so as to face the bandpass filter and the surface electrode, the incident light is multiplexed between them. While repeating the reflection, the inside of the light absorbing member is diffused in a direction parallel to the light receiving surface. Therefore, the absorption rate of incident light by the light absorbing member is improved. In particular, when the diffuse reflection surface satisfies the formula (1) described later, the absorption rate of incident light by the light absorbing member is further improved.

Further, when the thickness of the light diffusion portion is set to a predetermined thickness, a photoelectric conversion element having a small series resistance can be formed.
Furthermore, when at least one of the front surface electrode and the back surface electrode has a laminated structure of the first layer and the second layer, formation of a Schottky bond can be avoided without reducing the light absorption rate.

It is a schematic block diagram of the photoelectric conversion element which concerns on one embodiment of this invention. It is a figure which shows the composition range of the acid etching liquid suitable for the roughening or curved surface of a pyramid shape tip part and a side surface. It is a cross-sectional schematic diagram of the non-reflective surface shape (texture) provided in the light-receiving surface side of the conventional solar cell. It is a cross-sectional schematic diagram of the diffuse reflection surface provided in the back surface side of the photoelectric conversion element.

It is an energy band figure of the solar cell in which the 1st diffused layer exists in the light-receiving surface vicinity. It is an energy band figure of the interface of silver and silicon (second diffusion layer). It is a figure which shows the relationship between the density | concentration at the time of etching p-type single crystal silicon with the tetramethylammonium hydroxide aqueous solution (liquid temperature about 90 degreeC), and an etching rate. It is a scanning electron microscope (SEM) photograph of the etched surface of p-type single crystal silicon etched for 3 hours at an etching rate of 50 to 60 μm / h using a tetramethylammonium hydroxide aqueous solution.

It is a SEM photograph of Si surface etched for 90 seconds with acid etching liquid. The acid composition in the acid etchant is (a) HF: HNO ₃ : H ₂ SO ₄ = 4: 16: 80, (b) HF: HNO ₃ : H ₂ SO ₄ = 10: 15: 75, (c ) HF: HNO ₃ : CH ₃ COOH = 18: 36: 46. (A) 11 wt% TMAH aqueous solution, (b) 11 wt% TMAH aqueous solution, and 15 wt% TMAH aqueous solution containing 0.05 wt% polyethylene glycol mono-4-octylphenyl ether, (c) 11 % By weight TMAH aqueous solution and HF: HNO ₃ : CH ₃ COOH = 20: 40: 40 acid etching solution, or (d) 11% by weight TMAH aqueous solution, and HF: HNO ₃ : H ₂ SO ₄ = 4 : SEM photographs of Si surfaces etched with an acid etchant of 16:80, respectively. FIG. 11A shows a case where light having a wavelength of 1064 nm is incident on a thin silicon substrate (sample No. 21) and a silicon substrate (sample No. 22 to sample No. 26) having an uneven structure on the surface. The reflection angle dependence of the standardized reflection intensity (I / I ₀ ) at the time. FIG. 11B is a diagram showing the relationship between the reflection angle θ and (I / I _o ) · sin θ. When it is assumed that the angular distribution of reflectance is cos ⁿ θ, the relationship between the n value and the full width at half maximum (FIG. 12A), and the relationship between the full width at half maximum and the light absorption rate (FIG. 12B). )). FIG. 13A shows current-voltage characteristics when the second diffusion layer is formed on the p-type single crystal silicon substrate and the silver / aluminum stacked electrode is formed on the second diffusion layer. FIGS. 13B and 13C are transmission electron micrographs of a cross section of a p-type single crystal silicon substrate on which a silver / aluminum laminated electrode having an aluminum thickness of 1.8 nm or 2.7 nm is formed, respectively. is there.

It is a figure which shows the relationship between the thickness of the aluminum layer of a silver / aluminum laminated electrode, and a reflectance when the light of wavelength 1064nm advances toward a silver / aluminum laminated electrode from a 2nd diffused layer. It is a figure which shows the relationship between the distance (propagation distance) from the incident position of light, and the light intensity at the point. It is a figure which shows the relationship between the distance from the light beam center of diameter 50 micrometers (FIG. 16 (a)) or diameter 100 micrometers (FIG.16 (b)), and the light intensity at the point. It is a figure which shows the relationship between the thickness of Si wafer, and conversion efficiency. It is a figure which shows the relationship between the radius of an etching part, and conversion efficiency when the light beam of diameter 50 micrometers (FIG. 18 (a)) or diameter 100 micrometers (FIG.18 (b)) is made to inject into Si with which the back surface was etched. .

Hereinafter, an embodiment of the present invention will be described in detail.
[1. Photoelectric conversion element]
The photoelectric conversion element according to the present invention has the following configuration.
(1) The photoelectric conversion element includes a light-absorbing member made of a light-absorbing material capable of absorbing monochromatic light having a wavelength λ ₀ and generating carriers, and substantially parallel rays made of the monochromatic light as incident light. Is used.
(2) The photoelectric conversion element is
The light absorbing member;
A band pass filter formed on the light receiving surface side surface of the light absorbing member;
A surface electrode formed around the bandpass filter;
A surface on the back side of the light absorbing member, a diffuse reflection surface formed at a position facing the band pass filter and the surface electrode;
A back electrode formed so as to cover all or part of the surface of the diffuse reflection surface.
(3) The band-pass filter has a function of selectively transmitting at least light having the wavelength λ ₀ .
(4) The diffuse reflection surface has a function of diffusing the incident light in a direction parallel to the light receiving surface by multiple reflection.
(5) The front electrode and the back electrode each have a function as a current collector for extracting the carriers and a function as a reflective film for reflecting the incident light.

[1.1. Incident light]
In the present invention, the incident light is composed of monochromatic light having a wavelength λ ₀ . That is, the photoelectric conversion element according to the present invention is a photoelectric conversion element for monochromatic light irradiation.
“Monochromatic light of wavelength λ ₀ ” means that the center wavelength is λ ₀ and the ratio of the half-width (Δλ / 2) of the spectrum to the center wavelength λ ₀ (= Δλ / 2λ ₀ ) is 0.04 or less. A certain light.

In the present invention, the incident light consists of substantially parallel light rays that are incident substantially in parallel toward the light receiving surface (the surface on which the incident light is incident).
The “substantially parallel light beam” refers to light having an angular distribution β indicating a traveling direction of light of 23 ° or less. The smaller β is, the better. In order to obtain high conversion efficiency, the incident light is preferably a parallel ray with β≈0.

In the present invention, incident light enters the light receiving surface at an incident angle θ.
The “incident angle θ” refers to an angle formed by the normal direction of the light receiving surface and the incident direction of incident light.
The “incident light incident direction” refers to the average traveling direction of incident light. For example, when the incident light is a parallel light beam (β = 0), the incident direction is a direction parallel to the incident light. Further, for example, when the traveling direction of incident light is in a conical shape with apex angle: 2β (β ≠ 0), the incident direction is a direction parallel to the central axis of the cone.
In the present invention, the incident angle θ may be greater than zero (= normal incidence) as long as the desired conversion efficiency is obtained. However, if the incident angle θ becomes too large, the conversion efficiency decreases. Therefore, the incident angle θ is preferably 23 ° or less. The incident angle θ is more preferably 5 ° or less.

[1.2. Light absorbing material]
The photoelectric conversion element includes a light-absorbing member made of a light-absorbing material that can absorb monochromatic light having a wavelength λ ₀ and generate carriers.
In the present invention, the light-absorbing material can be either a direct transition type semiconductor or an indirect transition type semiconductor, and may be appropriately selected according to the wavelength of incident light. Examples of the light absorbing material include Si, Ge, GaP, and GaAs.

The composition of the light absorbing material is determined by the wavelength λ ₀ (nm) of monochromatic light (the energy of the light E _L [eV] = hc / eλ ₀ = 1240 / λ ₀ ) and the forbidden band width E _g of the light absorbing material. .
For example, in the case of λ ₀ = 1064 nm, the light absorbing material is preferably silicon, particularly single crystal silicon. In order for light with energy E _L to be absorbed by the semiconductor having the forbidden bandwidth E _g , it is necessary that E _g <E _L. When light is absorbed by such a semiconductor, the energy of E _L -E _g becomes heat and does not contribute to power generation, so that E _L and E _g are preferably close. The energy E _L of light at λ ₀ = 1064 nm is 1.17 eV, and the forbidden band width E _g of silicon is 1.11 eV. Both are preferred combinations because E _L and E _g are close.
Similarly, for light with λ ₀ = 800 to 860 nm (E _L = 1.55 to 1.44 eV), for example, gallium arsenide with E _g = 1.43 eV is suitable.

The conductivity type of the light-absorbing material constituting the main part of the light-absorbing member may be either p-type or n-type, and is optimal depending on the composition of the light-absorbing material and the wavelength λ _{0 of} monochromatic light Select.
When the light-absorbing material absorbs light, minority carriers (electrons in the case of p-type and holes in the case of n-type) are generated. As described in Non-Patent Document 3, the diffusion length of minority carriers is longer for electrons than for holes.
On the other hand, in the case of a single crystal silicon substrate, the thickness of the first diffusion layer is generally set to 1 μm or less. When the thickness of the first diffusion layer is compared with the thickness of the other portions, the latter is overwhelmingly thick.
Therefore, when single crystal silicon is used for the light absorbing material, it is preferable to select the p-type so that minority carriers are electrons.

[1.3. Light absorbing member]
A band pass filter and a surface electrode, which will be described later, are formed on the surface of the light absorbing member on the light receiving surface side. On the other hand, a diffuse reflection surface described later is formed on the surface opposite to the light receiving surface, and a back electrode is formed on the surface of the diffuse reflection surface.
A first diffusion layer (emitter layer) having a conductivity type different from that of the light absorbing member is formed on one surface of the light absorbing member. Furthermore, a second diffusion layer (back surface field (BSF) layer) showing the same conductivity type as that of the light absorbing member may be provided on the other surface of the light absorbing member, if necessary.

[1.3.1. Light diffusion thickness]
In the photoelectric conversion element according to the present invention, incident light is absorbed by the light absorbing member while repeating multiple reflections between the front surface and the back surface of the light absorbing member. Therefore, the thickness of the light diffusion portion and the penetration depth of incident light into the light absorbing material affect the conversion efficiency.

Here, the “light diffusing portion” means that when the light absorbing member is viewed from the normal direction of the light receiving surface, the projection surface of the bandpass filter and the surface electrode and the projection surface of the diffuse reflection surface are An overlapping part. Incident light repeats multiple reflections at least between the bandpass filter and the surface electrode and the diffuse reflection surface. In the case where the back electrode is formed in a portion of the diffuse reflection surface that contributes to multiple reflection, the back electrode also contributes to multiple reflection. In order to obtain high conversion efficiency, it is preferable to cover the entire surface of the diffuse reflection surface, which contributes to multiple reflection, with the back electrode, and to reflect light on both the diffuse reflection surface and the back electrode.
The “depth of penetration of incident light into the light absorbing material” is a depth at which the intensity of the incident light becomes 1 / e (e is the number of Napier) when the incident light is incident on the light absorbing material. Say.

In general, the greater the thickness of the light diffusion portion, the longer the distance traveled by the generated carriers and the higher the series resistance. On the other hand, if the thickness of the light diffusion portion becomes too thin, there is a problem that light cannot be sufficiently absorbed.
If the thickness of the light diffusion part is 1/200 or more of the penetration depth of the incident light into the light absorbing material due to the multiple reflection effect by the band pass filter on the light receiving surface side and the diffuse reflection surface on the back surface side, the incident light is It can be absorbed at a considerable rate. When the thickness is further increased, the light absorption rate is gradually improved. However, when the thickness becomes 1/10 of the penetration depth of the incident light into the light absorbing material, the incident light is almost completely absorbed. Therefore, if it is thicker than that, only the adverse effect of increasing the series resistance increases, and as a result, the conversion efficiency decreases.
Therefore, the thickness of the light diffusion portion is preferably 1/200 to 1/10 of the penetration depth of incident light into the light absorbing material.

  For example, when the light absorbing member is single crystal silicon, the thickness of the light diffusion portion is preferably 100 μm or less. The thickness of the light diffusion portion is more preferably 50 μm or less, and still more preferably 20 μm or less.

  The thickness of the part other than the light diffusion part is not particularly limited, and an optimum thickness can be selected according to the purpose. In the present invention, since the diffuse reflection surface (uneven structure) is formed on the back side of the light absorbing member, the mechanical strength of the light absorbing member may be reduced if the entire thickness of the light absorbing member is as described above. . Therefore, it is preferable to form a thick part thicker than the light diffusion part around the light diffusion part.

[1.3.2. Size of light diffuser]
In the present invention, the bandpass filter and the front surface electrode, and the diffuse reflection surface and the back surface electrode all have a function of reflecting incident light or a function of rereflecting the reflected light inside the light absorbing member. Incident light diffuses in a direction parallel to the light receiving surface of the light absorbing member while repeating multiple reflections between the front and back surfaces of the light diffusing section. Therefore, the size of the light diffusion portion (that is, the size of the front electrode, the back electrode, and the diffuse reflection surface) affects the conversion efficiency.

Note that the size of the diffuse reflection surface does not necessarily need to completely match the size of the band-pass filter + surface electrode. This is because the overlapping part of both projection surfaces becomes a light diffusion part.
Similarly, the size of the diffuse reflection surface does not necessarily need to completely match the size of the back electrode. However, it is difficult to completely reflect the incident light only by the diffuse reflection surface. Therefore, it is preferable that the entire surface of the diffuse reflection surface that contributes to multiple reflection is covered with the back electrode.

When light diffuses to a region outside the light diffusion portion (for example, a region where only the surface electrode exists), most of the light is transmitted to the outside of the light absorbing member as it is without being re-reflected. Therefore, the amount of light absorbed by the light absorbing member decreases as the size of the light diffusion portion decreases.
In order to obtain high conversion efficiency, the shortest distance from the edge of the bandpass filter to the edge of the light diffusing portion is preferably at least twice the diffusion distance. More preferably, the shortest distance is at least three times the diffusion distance.
Here, the “diffusion distance” means that the band-pass filter and the diffusive reflection surface are formed on the front and back surfaces of the light-absorbing material having a thickness corresponding to the thickness of the light diffusion portion, respectively. When the incident light is diffused in a direction parallel to the light receiving surface while multiple reflection of the incident light is performed between a band pass filter and the diffuse reflection surface, the intensity of the incident light is 1 / e (e is , Napier number).

As the size of the light diffusion portion increases, more multiple reflections are repeated, so that the conversion efficiency improves. However, the effect is saturated as the size of the light diffusion portion increases. Therefore, even if the size of the light diffusion portion is increased more than necessary, there is no practical benefit.
Therefore, the shortest distance from the edge of the bandpass filter to the edge of the light diffusion portion is preferably 10 times or less of the diffusion distance. The shortest distance is more preferably 5 times or less of the diffusion distance, and further preferably 4 times or less of the diffusion distance.

[1.3.3. Diffusion layer]
A first diffusion layer (emitter layer) having a conductivity type different from that of the light absorbing member is formed on either the back surface side or the light receiving surface side of the light absorbing member. If necessary, a second diffusion layer (BSF layer) showing the same conductivity type as that of the light absorbing member may be formed on the other surface. Which surface the emitter layer is provided on can be selected according to the composition of the light absorbing material, the method of forming the diffusion layer, and the like.

In order to obtain high conversion efficiency, the thickness of the diffusion layer is preferably uniform. However, depending on the composition of the light absorbing member and the type of dopant, it may be difficult to form a uniform diffusion layer on the uneven surface. For example, when the light absorbing member is p-type single crystal silicon, it is difficult to form a BSF layer having a uniform thickness on the uneven surface. Therefore, in such a case, it is preferable to form an emitter layer on the back surface of the light absorbing member.
When the emitter layer is formed on the back surface of the light absorbing member, the incident light may be attenuated before the incident light reaches the emitter layer. However, the attenuation of the incident light can be reduced by reducing the thickness of the light diffusion portion.

[1.4. Band pass filter]
[1.4.1. Definition]
In the photoelectric conversion element according to the present invention, a band pass filter is formed on a surface (light receiving surface) on which the monochromatic light is incident at an incident angle θ.
“Band pass filter” refers to a filter having a function of selectively transmitting at least light of wavelength λ ₀ . The bandpass filter is usually used to extract light having a specific wavelength from white light. In the present invention, the band-pass filter is used to transmit incident light (monochromatic light) as it is and to reflect reflected light from the back surface into the light absorbing member again. This is different from the conventional one.

[1.4.2. Center wavelength λ]
The wavelength of light that can be transmitted through the band-pass filter has a predetermined width (transmission band).
“Center wavelength λ” refers to the median value of the wavelength of light that can pass through the band-pass filter.
The phrase “at least selectively transmits light having the wavelength λ ₀ ” means that the center wavelength λ of the bandpass filter does not necessarily need to completely match the center wavelength λ ₀ of the incident light. However, if the difference between the wavelength λ and the wavelength λ ₀ is greatly different, the number of photons that pass through the band-pass filter is reduced, and the conversion efficiency is lowered.
In order to obtain high conversion efficiency, the center wavelength λ of the bandpass filter is preferably 0.97 × λ ₀ or more and 1.039 × λ ₀ or less. The center wavelength λ is more preferably 0.98 × λ ₀ or more and 1.025 × λ ₀ or less, and further preferably 0.99 × λ ₀ or more and 1.01 × λ ₀ or less.

[1.4.3. Construction]
The structure of the bandpass filter is not particularly limited as long as it exhibits the above-described function. As a bandpass filter, for example,
(A) (L4 / H4 /) ^m1 L2 (H4 / L4 /) ^m2 laminated structure,
(B) having a laminated structure of AR / (L4 / H4 /) ^m1 L2 (H4 / L4 /) ^m2 ,
(C) (H4 / L4 /) ^m1 H2 (L4 / H4 /) ^m2 laminated structure,
(D) AR / (H4 / L4 /) ^m1 H2 (L4 / H4 /) ^m2 having a laminated structure,
and so on.

However,
“AR” is an antireflection layer made of a material having a refractive index of n _AR ,
“L4” is a low refractive index layer (A) made of a low refractive material (A) having a refractive index of n _L4 ,
“L2” is a low refractive index layer (B) made of a low refractive material (B) having a refractive index of n _L2 ;
“H4” is a high refractive index layer (A) made of a high refractive material (A) having a refractive index of n _H4 (> n _L4 ,> n _L2 ),
“H2” is a high refractive index layer (B) made of a high refractive material (B) having a refractive index of n _H2 (> n _L4 ,> n _L2 ),
m ₁ and m ₂ are each an integer of 1 or more.

"L4", "L2", respectively, the optical thickness (= the refractive index n × actual thickness d) represents a layer that is proportional to lambda _0/4 or λ _0/2. If the incident angle theta, conditions of the optical thickness nd of each layer constituting the band-pass filter, nd = a ^{(1-sin 2 θ / n} 2) 1/2 × (λ 0/4 or λ 0/2) Become. This also applies to “H4” and “H2”.
“(L4 / H4 /) ^m1 ” represents that the lamination unit of “L4 / H4 /” is repeated m ₁ times. The number of repetitions m _{1 in the first} half and the number of repetitions m ₂ in the _second half may be the same or different.
“(L4 / H4 /) ^m1 L2 (H4 / L4 /) ^m2 ” means that the L2 layer is inserted between the stacked structure of (L4 / H4 /) ^{m1 and} the stacked structure of (H4 / L4 /) ^m2. Represents that
“AR /” indicates that an antireflection layer is formed on the outermost surface (light incident side) of the bandpass filter.

The low refractive index material (A) constituting L4 and the low refractive index material (B) constituting L2 may be the same material or different materials as long as the conditions described later are satisfied. . Further, the individual L4 included in the laminated structure may be the same material or different materials as long as the conditions described later are satisfied.
Similarly, the high refractive index material (A) constituting H4 and the high refractive index material (B) constituting H2 may be the same material or different materials as long as the conditions described later are satisfied. There may be. Further, the individual H4 contained in the laminated structure may be the same material or different materials as long as the conditions described later are satisfied.

  The band pass filter is particularly preferably provided with a laminated structure (a) or a laminated structure (b). A band-pass filter having such a laminated structure can obtain higher conversion efficiency than a band-pass filter having another laminated structure.

An ideal thickness (actual thickness) d ₀ exists for the thickness (actual thickness) d of each layer constituting the bandpass filter. Of each layer thickness d is not necessarily completely identical with the ideal thickness d _0. However, if the difference between the thickness d and the ideal thickness d ₀ increases, the conversion efficiency decreases. Thus, the thickness d is closer to the ideal thickness d ₀ is preferable.

For example, in the case of a bandpass filter having the above-described laminated structure (a) or laminated structure (b), the thickness d _{L4 of} the low refractive index layer (A), the thickness d _{L2 of} the low refractive index layer (B), The thickness d _H4 of the high refractive index layer (A) preferably satisfies the following relationship.

That is, the thickness d _L4 of the low refractive index layer (A) is preferably 0.75 × d _L40 or more and 1.20 × d _L40 or less. The thickness d _L4 is more preferably 0.85 × d _L40 or more and 1.10 × d _L40 or less, and further preferably 0.92 × d _L40 or more and 1.06 × d _L40 or less.
However, d _L40 is an ideal thickness of the low refractive index layer (A) (actual _{thickness), d L40 = (1-} sin 2 θ / n L4 2) 1/2 × (λ 0/4 ) × (1 / n _L4 ).

The thickness d _L2 of the low refractive index layer (B) is preferably 0.926 × d _L20 or more and 1.058 × d _L20 or less. The thickness d _L2 is more preferably 0.95 × d _L20 or more and 1.03 × d _L20 or less, and further preferably 0.97 × d _L20 or more and 1.02 × d _L20 or less.
However, d _L20 is an ideal thickness of the low refractive index layer (B) (actual _{thickness), d L20 = (1-} sin 2 θ / n L2 2) 1/2 × (λ 0/2 ) × (1 / n _L2 ).

The thickness d _H4 of the high refractive index layer (A) is preferably 0.87 × d _H40 or more and 1.10 × d _H40 or less. The thickness d _H4 is more preferably 0.92 × d _H40 or more and 1.05 × d _H40 or less, and still more preferably 0.96 × d _H40 or more and 1.02 × d _H40 or less.
However, d _H40 is an ideal thickness of the high refractive index layer (A) (actual _{thickness), d H40 = (1-} sin 2 θ / n H4 2) 1/2 × (λ 0/4 ) × (1 / n _H4 ).

In the bandpass filter having the laminated structure (b) or the laminated structure (d), the refractive index n _AR and the thickness (actual thickness) d _AR of the antireflection layer are not particularly limited. That is, the antireflection layer may be a layer made of a material that transmits at least light.

Further, in the bandpass filter having the laminated structure (b), in order to obtain higher conversion efficiency, the refractive index n _AR and the thickness d _AR of the antireflection layer satisfy the following relationships, respectively. Is preferred.

That is, the refractive index n _AR of the antireflection layer is preferably 0.60 × n _pv ^1/2 or more. The refractive index n _AR is more preferably 0.7 × n _pv ^1/2 or more and 1.8 × n _pv ^1/2 or less, more preferably 0.8 × n _pv ^1/2 or more and 1.5 × n. _pv ^1/2 or less.
Where n _pv is the refractive index of the light absorbing material.

The thickness d _AR of the antireflection layer is preferably 1.70 × d _AR0 or less. The thickness d _AR is more preferably 0.4 × d _AR0 or more and 1.4 × d _AR0 or less, and further preferably 0.7 × d _AR0 or more and 1.2 × d _AR0 or less.
However, d _AR0 is an ideal thickness of the antireflection layer (actual _{thickness), d AR0 = (1-} sin 2 θ / n AR 2) 1/2 × (λ 0/4) × (1 / N _AR ).

[1.4.4. material]
The material of each layer constituting the band pass filter is not particularly limited, and various materials can be used depending on the purpose.
Examples of the low refractive index material include MgF ₂ (refractive index of light having a wavelength of 1064 nm: about 1.36), SiO ₂ (refractive index of light having a wavelength of 1064 nm: about 1.43), and the like.
Examples of the high refractive index material include ZnS (refractive index of light having a wavelength of 1064 nm: about 2.29), TiO ₂ (refractive index of light having a wavelength of 1064 nm: about 2.25), and the like.
Examples of the material for the antireflection layer include Y ₂ O ₃ (refractive index of light having a wavelength of 1064 nm: about 1.73), composite oxides such as TiO ₂ —SiO _2, and the like.

[1.4.5. Size of bandpass filter]
A surface electrode is formed around the bandpass filter. Therefore, if the size of the bandpass filter is smaller than the size of the cross section of the incident light, a part of the incident light is reflected by the surface electrode. Therefore, the size of the band pass filter is preferably equal to or larger than the size of the cross section of the incident light.
On the other hand, even if the size of the bandpass filter is increased more than necessary, there is no practical benefit. What is necessary is just to design suitably in view of the ease of preparation of the thin film which comprises a band pass filter.

[1.5. Diffuse reflective surface]
The light absorbing member includes a diffuse reflection surface on the back surface. The diffuse reflection surface is for reflecting incident light that could not be absorbed by the light absorbing member. In other words, the diffuse reflection surface has a function of diffusing incident light in a direction parallel to the light receiving surface by multiple reflection. The diffuse reflection surface may be formed on the entire back surface of the light absorbing member, but may be formed at least in a portion that substantially contributes to multiple reflection of light (that is, a light diffusion portion).

In order to diffuse the incident light in a direction parallel to the light receiving surface, the diffuse reflection surface preferably has a reflection spectrum of the incident light satisfying the relationship of the following expression (1).
Full width at half maximum of (I / I ₀ ) · sinθ ≧ 30 ° (1)
Where I / I ₀ is the normalized reflection intensity, and θ is the reflection angle measured from the incident direction.
In order to obtain high conversion efficiency, the full width at half maximum is more preferably 50 ° or more.

As a diffuse reflection surface satisfying such conditions, for example,
(A) a surface on which pyramids with random sizes (the size and / or height of the bottom surface) are randomly arranged;
(B) a surface in which pyramids with random sizes are randomly arranged, and the tip and side surfaces of the pyramid are roughened or curved;
(C) a surface on which a wedge-shaped depression having a diameter of 1 to 5 μm is formed densely;
and so on.
According to a conventional diffused reflection surface fabrication evaluation report, there is an example in which the value of w is obtained by associating with the Phong model that approximates the reflection angle dependency of the reflection intensity by cos ^w θ, and this is used as an index of diffuse reflection. There, w = 10 (reference document A) is reported for the mirror silicon / screen printed Al interface, and w = 30 (reference document B) is reported for the alkali-etched silicon / deposited Ag interface. w = 10 and 30 correspond to the full width at half maximum of (I / I ₀ ) · sin θ of 26.2 ° and 16.2 °, respectively. Therefore, the conventional manufacturing method cannot provide a sufficient light diffusion reflection function.
On the other hand, a diffuse reflection surface satisfying the above conditions can be formed by an anisotropic etching method described later, an isotropic etching method using an acid, or a combination thereof.
[Reference A] M. Hermle, E. Schneiderlochner, and G. Grupp, SW Glunz, Proc. 20th European Photovoltaice Solar Energy Conference (Barcelona, Spain, 2005) 810
[Reference B] D. Kray, M. Hermle, and SW Glunz, Prog. Photovolt .: Res. Appl. 16, 1 (2008)

[1.6. Front electrode and back electrode]
A surface electrode is formed on the light receiving surface side of the light absorbing member and around the band pass filter. A back electrode is formed on the back surface of the light absorbing member at a position facing the bandpass filter and the front electrode (that is, the surface of the diffuse reflection surface). The front electrode and the back electrode each have a function as a current collector for extracting carriers and a function as a reflective film for reflecting incident light.

[1.6.1. Size and shape of electrode]
The size and shape of the front surface electrode and the back surface electrode are not particularly limited, and an optimal one can be selected according to the purpose. Further, the front electrode and the back electrode need not have the same size and / or shape. Of the front electrode and the back electrode, a portion that contributes to multiple reflection of light functions as a light diffusion portion. Therefore, it is preferable to select the size and shape of the front and back electrodes so that high conversion efficiency can be obtained.
In order to obtain high conversion efficiency, the back electrode preferably covers the entire surface of the diffuse reflection surface. However, as long as the conversion efficiency is not significantly reduced, the back electrode may cover a part of the diffuse reflection surface.

[1.6.2. Electrode material]
A metal material is usually used for the electrode. When a metal electrode is formed directly on the surface of the light absorbing member, a Schottky junction may be formed between the light absorbing member / electrode. When a Schottky junction is formed between the light absorbing member / electrode, high conversion efficiency cannot be obtained. Therefore, the electrode material is preferably a material capable of forming an ohmic junction with the light absorbing material.
In the present invention, since the electrode also functions as a reflective film, the electrode is required to have a high reflectance with respect to incident light. Specifically, the reflectance is preferably 90% or more. The reflectance is more preferably 95% or more.

When there is a material that satisfies both of these two conditions for a certain light-absorbing material, an electrode can be formed directly on the surface of the light-absorbing material.
On the other hand, if there is no material that satisfies these two conditions at the same time, the electrode
A first layer made of a conductive material provided on a surface of the light absorbing member and capable of forming an ohmic junction with the light absorbing member;
It is preferable to have a laminated structure with a second layer made of a second conductive material provided on the surface of the first layer and having a higher reflectance of the incident light than the first conductive material.
An electrode having such a laminated structure may be used for either the front electrode or the back electrode, or may be used for both.

In the electrode having a stacked structure, carriers are taken out from the first layer through the second layer. In the case where the second layer is a material that forms a Schottky junction with the light-absorbing material, if the thickness of the first layer is too thin, the carriers are affected by the second layer. In order to obtain high conversion efficiency, the thickness of the first layer is preferably at least three times the penetration depth of carriers into the first conductive material. The thickness of the first layer is more preferably 5 times or more the penetration depth of the carrier.
Here, “the penetration depth of the carrier into the first conductive material” is assumed that the carrier (A) in the light absorbing material and the carrier (B) in the first conductive material can be distinguished, When the concentration distribution in the first conductive material of the carrier (A) in the light absorbing material is calculated using Poisson's equation, the concentration of the carrier (A) in the first conductive material is The depth which becomes 1 / e (e is the Napier number) of the concentration of the carrier (A) in the absorbent material.

Part of the incident light is transmitted through the diffuse reflection surface. In the electrode having the laminated structure, part of the transmitted light further passes through the first layer and is reflected by the surface of the second layer. If the thickness of the first layer is too thick, the attenuation of light in the first layer increases. In order to obtain high conversion efficiency, the thickness of the first layer is preferably equal to or less than the penetration depth of incident light into the first conductive material. More preferably, the thickness of the first layer is not more than ½ times the penetration depth of incident light.
Here, “the penetration depth of incident light into the first conductive material” means that the incident light intensity is 1 / e (e is the Napier number) when the incident light is incident on the first conductive material. ).

For example, the incident light is monochromatic light with λ ₀ = 1064 nm, the light-absorbing material is p-type single crystal silicon, an emitter layer (n-type) is formed on the back surface, and the BSF layer is formed on the light receiving surface. Consider the case of forming (p-type).
For example, Ag is known as a material having a high reflectance. Ag hardly forms a Schottky junction with the emitter layer of p-type single crystal silicon. Therefore, Ag can be used as it is for the back electrode.

On the other hand, when Ag is used as the surface electrode, a Schottky junction is formed when Ag is brought into direct contact with the BSF layer of p-type single crystal silicon.
On the other hand, Al is a general electrode material used for silicon LSI. When an Al layer is formed on the silicon surface and then a heat treatment (sintering process) is performed at a temperature of about 400 ° C. for about several tens of minutes, Si diffuses into the Al. The sintered Al forms an ohmic junction with both the p-type and n-type conductive layers. However, the reflectance of Al is lower than Ag.

  The photoelectric conversion element of the present invention increases the optical path length in the light absorbing member by performing multiple reflection between the light receiving surface side surface and the back surface side of the light absorbing member, and absorbs light. It is characterized by raising. For this reason, the fact that the reflectance of Al is slightly smaller than that of Ag has a decisive influence on the decrease in conversion efficiency. Therefore, it is preferable to use a laminated electrode of Al (first layer) and Ag (second layer) instead of using Al as it is for the surface electrode.

According to Reference Document 1 below, the extinction coefficient k of Al for light of λ ₀ = 1064 nm is k = 10.6, so the penetration depth of Al at a wavelength of 1064 nm is calculated to be about 8 nm. . Therefore, if the thickness of the Al layer is sufficiently thinner than 8 nm, it is as if there is no Al layer for light having a wavelength of 1064 nm. It will be reflected by.
[Reference 1] ED Palik, Handbook of Optical Constant of solids, (Academic Press 1985)

  On the other hand, the penetration depth of electrons which are minority carriers in the BSF layer (p-type) is about 0.1 nm. Therefore, if the thickness of the Al layer is sufficiently thicker than 0.1 nm, electrons feel that the Al layer is in contact with the BSF layer. Therefore, a stacked electrode of an Al layer (first layer) and an Ag layer (second layer) satisfying such conditions may be formed on the BSF layer.

  However, in order to make ohmic contact between the Al layer and the BSF layer, it is necessary to perform a heat treatment for several tens of minutes at a temperature of about 400 ° C. Since the Al layer is overwhelmingly thinner than the Ag layer, Al and Ag are alloyed by a heat treatment of about several tens of minutes at a temperature of about 400 ° C., and the probability that Ag atoms are exposed on the surface of the Al layer increases. . In order to prevent this from happening, it is necessary to determine the temperature and time of the sintering process.

[1.7. Concrete example]
In FIG. 1, the schematic block diagram of the photoelectric conversion element which concerns on one embodiment of this invention is shown. In FIG. 1, the photoelectric conversion element 10 includes a light absorbing member 12, a band pass filter (BPF) 14, a front surface electrode 16, and a back surface electrode 18.

The light absorbing member 12 is made of a p-type semiconductor substrate (for example, a p-type single crystal silicon substrate) having a predetermined thickness. A BPF 14 is formed on the surface of the light absorbing member 12 on the light receiving surface side through the SiO ₂ film 20. The SiO ₂ film has a thickness of about 13 nm and protects the surface of the light absorbing member (silicon) 12. In order to exhibit this protective function, a thickness of about 13 nm is required. However, at this time, it is necessary to adjust the thickness of the low refractive index layer in contact with the light absorbing member 12 among the layers constituting the BPF 14. A surface electrode 16 is formed on the light receiving surface side of the light absorbing member 12 and around the BPF 14. The surface electrode 16 is a laminated electrode of a first layer 16a made of Al and a second layer 16b made of Ag. Further, a BSF layer (second diffusion layer) 12a is formed immediately below the surface on the light receiving surface side of the light absorbing member.

A thin plate portion is formed at the center on the back surface side of the light absorbing member 12, and a diffuse reflection surface 12b is formed on the entire surface of the thin plate portion. An emitter layer (first diffusion layer) 12c is formed immediately below the surface on the back surface side of the light absorbing member 12 including the diffuse reflection surface 12b. Further, a back electrode (for example, an Ag electrode) 18 is formed on the back surface of the light absorbing member 12 including the diffuse reflection surface 12b.
The thin plate portion of the light absorbing member 12 is a “light diffusing portion”, that is, when the light absorbing member 12 is viewed from the normal direction of the light receiving surface, the projection surface of the bandpass filter 14 and the surface electrode 16, and the diffuse reflecting surface 12b. This corresponds to the portion where the projection plane overlaps.

When monochromatic light having a wavelength λ ₀ is incident on the BPF 14 of the photoelectric conversion element 10 shown in FIG. 1, the monochromatic light is transmitted through the BPF 14 and reflected by the diffuse reflection surface 12b. Part of the incident light passes through the diffuse reflection surface 12 b and is reflected by the back electrode 18. The light reflected by the diffuse reflection surface 12 b and the back surface electrode 18 is further reflected again by the BPF 14 or the front surface electrode 16. Hereinafter, the light absorption member 12 absorbs light energy while repeating such multiple reflection.

  At this time, if the shape of the diffusive reflecting surface 12b, the thickness and size of the light diffusing portion, and the like are optimized, the light absorption rate is improved by the light confinement effect and the recombination of carriers is suppressed. Further, if the materials of the front electrode 16 and the back electrode 18 are optimized, the formation of a Schottky junction can be avoided. As a result, higher conversion efficiency can be obtained as compared with conventional photoelectric conversion elements.

[2. Method for manufacturing photoelectric conversion element]
The photoelectric conversion element according to the present invention is
(A) If necessary, form a thin plate portion to be a light diffusion portion in the light absorbing member,
(B) forming a diffuse reflection surface at least on the surface on the back side of the light diffusion portion (thin plate portion);
(C) forming a diffusion layer (an emitter layer and, if necessary, a BSF layer) on the surface of the light absorbing member;
(D) A band pass filter and a surface electrode are formed on the light receiving surface side surface of the light absorbing member, and a back surface electrode is formed on the back surface side of the light absorbing member.
Can be manufactured.
In addition, the order of each process is not limited to said order, A procedure can be made to move back and forth as much as physically possible.

[2.1. Thin plate part forming process]
When the light absorbing member is a bulk material such as a single crystal silicon substrate, a thin plate portion serving as a light diffusion portion is formed on the back surface side of the light absorbing member (thin plate portion forming step). When the light absorbing member has a thickness suitable for multiple reflection of light from the beginning (for example, when the light absorbing member is a thin film), the thin plate portion forming step can be omitted.

[2.1.1. Method for forming thin plate portion]
As a method of forming the thin plate portion, for example,
(A) A method of masking a region other than the region to be thinned out of the back surface of the light absorbing member and immersing the masked light absorbing member in an etching solution,
(B) a method of etching the entire back surface of the light absorbing member without masking;
(C) There is a method of removing the entire back surface of the light absorbing member by machining such as polishing.
When the light absorbing member is a bulk material, when the entire light absorbing member is thinned, the strength of the light absorbing member is lowered, which is not practical. Therefore, it is preferable to use an etching method to thin only the portion that substantially contributes to photoelectric conversion.

[2.1.2. Etching solution for forming thin plate portion]
When thinning is performed by etching, the etching solution is not particularly limited, and an optimum one is selected according to the composition of the light absorbing member. However, when performing masking, it is necessary to use an etching solution that does not peel off or disappear the masking material.
As an etchant, for example,
(A) an alkaline etching solution containing sodium hydroxide, potassium hydroxide, tetramethylammonium hydroxide, ethylenediamine-pyrocatechol, hydrated hydrazine, etc.
(B) an acid etching solution mainly composed of hydrofluoric acid and nitric acid;
and so on.
At the time of thinning, any one etching solution may be used, or two or more etching solutions may be used in combination.

Among these, the alkaline etching solution has the following characteristics.
(A) Generally, since anisotropic etching is performed, it is easy to form an uneven etched surface. However, the etched surface may be flat depending on the etching conditions.
(B) As a mask material, a silicon oxide film that can easily form and remove a mask pattern can be used.
(C) The etching rate is slower than that of the acid etching solution.

On the other hand, the acid etching solution has the following characteristics.
(A) Since it is generally isotropic etching, it is easy to form a flat etching surface. However, when there is a difference in microscopic crystal structure on the surface of the light absorbing member, or when the composition of the acid etching solution is optimized, an uneven etched surface can be formed.
(B) A silicon oxide film may not be used as a mask material.
(C) The etching rate is faster than that of the alkaline etching solution.

  In the etching for thinning, if the etching rate is too slow, the etching time is not only long, but it is not practical for controlling the liquid temperature and concentration. For example, when the light absorbing member is single crystal silicon, the thickness (practical initial thickness) that has sufficient mechanical strength and can complete the etching process within a practical time is 150 μm to 300 μm. The final thin plate portion has a thickness of 5 to 50 μm. Considering these, the etching rate of the etching solution is preferably 50 μm / h or more.

  When the composition of the acid etching solution is optimized, an etching rate of silicon of 300 μm / h to 1200 μm / h is obtained. Therefore, in consideration of the practical initial thickness and the final thickness of the thin plate portion, when etching is performed using an acid etching solution, a thin plate portion having the desired thickness can be obtained in about 10 minutes. It will be.

However, in the case where silicon is acid-etched, if a silicon oxide film is used as a mask material, even if the composition of the acid etching solution is optimized, the silicon oxide film is corroded by the acid etching solution and is masked within 10 minutes. The effect cannot be demonstrated.
In order to prevent this, for example, a silicon nitride film may be used as the mask material. However, formation and removal of a mask pattern using a silicon nitride film is very complicated.
Therefore, in the case where the light absorbing member is silicon, it is preferable to use an alkaline etching solution when partially thinning the light absorbing member.

For alkaline etchants,
(A) sodium hydroxide aqueous solution, potassium hydroxide aqueous solution,
(B) an aqueous tetramethylammonium hydroxide (TMAH) solution,
(C) Ethylenediamine-pyrocatechol, hydrated hydrazine and the like can be used.

In the semiconductor process, the semiconductor substrate, various thin films stacked thereon, the processing liquid for processing the semiconductor substrate and the thin film, and the parts that are in contact with the semiconductor substrate and the thin film in the parts for performing various processing, such as sodium and potassium Alkaline metal ions or alkaline earth metal ions such as calcium are present at a concentration of about 0.001 ppb (10 ¹⁰ ions / cm ^{3 in} terms of the number density of ions). There is a case.
Therefore, it is preferable to use an alkaline etching solution which does not contain any of alkali metal ions and alkaline earth metal ions for etching for thinning. When an etching solution containing alkali metal ions or alkaline earth metal ions is used, an extremely strict cleaning process is required after the etching.

  Neither ethylenediamine-pyrocatechol nor hydrated hydrazine contains alkali metals and alkaline earth metals. However, when silicon is etched using these etching solutions, the temperature of the solution needs to be raised to 100 ° C. or higher in order to obtain an etching rate of 50 μm / h. Therefore, the etching facility becomes complicated.

  On the other hand, the TMAH aqueous solution does not contain alkali metal and alkaline earth metal ions. Further, an etching rate of 50 μm / h or more can be obtained at a temperature of about 90 ° C., and various etching rates can be adjusted by changing the TMAH concentration. Therefore, the TMAH aqueous solution is most suitable as an etching solution for thinning. In this case, the weight concentration of TMAH in the TMAH aqueous solution is preferably 13 to 22% by weight.

[2.2. Diffuse reflecting surface forming process]
Next, a diffuse reflection surface is formed at least on the surface on the back surface side of the light diffusion portion (thin plate portion) (diffuse reflection surface forming step).
The surface of the thinned part is usually smooth. Therefore, in order to make the surface of the thin plate portion a diffuse reflection surface, it is necessary to form random irregularities on at least the surface on the back surface side of the light diffusion portion. The method for forming the diffuse reflection surface is not particularly limited as long as the desired reflectance can be obtained. Specific examples of the method for forming the diffuse reflection surface include the following methods.

[2.2.1. Formation of diffuse reflection surface by acid etching]
A first specific example for forming the diffuse reflection surface is a method of roughening or curving at least the back surface of the light diffusion portion with an acid etching solution.
In many cases, the acid etching solution is isotropic etching, but when the composition is optimized, the surface of the light absorbing member can be roughened or curved.
As an acid etching solution used for roughening or curved surface,
(A) a mixed acid of hydrofluoric acid, nitric acid, and sulfuric acid,
(B) There is a mixed acid of hydrofluoric acid, nitric acid, and acetic acid.
In addition, the suitable composition of the mixed acid for forming a diffuse reflection surface is the same as that of the 2nd etching liquid mentioned later (refer FIG. 2).

[2.2.2. Random pyramid formation and rough or curved pyramid surface]
The second specific example for forming the diffuse reflection surface is:
(A) Using the first etching solution, a random pyramid is formed on the back surface of the light absorbing member;
(B) A method of roughening or curving the tip and side surfaces of the pyramid using a second etching solution.

  The first etching solution is preferably the alkaline etching solution described above. Since the alkaline etching solution is often anisotropic etching, a random pyramid can be easily formed. Among the alkaline etching solutions, an alkaline etching solution containing neither alkali metal ions nor alkaline earth metal ions is preferable, and a TMAH aqueous solution is particularly preferable.

  When a TMAH aqueous solution is used as the first etching solution, the weight concentration of TMAH is preferably 5 to 11% by weight. For example, when the light absorbing member is a single crystal silicon substrate, a random pyramid can be formed by etching for 1 to 2 hours with 5 to 11% by weight of TMAH aqueous solution.

As the second etching solution, for example,
(A) An alkaline etching solution containing neither alkali metal ions nor alkaline earth metal ions, wherein an additive having a chain molecular structure in which one end is hydrophobic and the other end is hydrophilic is added to the light. What adjusted the etching rate of the absorbing member to 15 μm / hour or less,
(B) an acid etching solution in which hydrofluoric acid, nitric acid and acetic acid are mixed, wherein the etching rate of the light absorbing member is adjusted to 10 μm / hour or less,
(C) An acid etching solution in which hydrofluoric acid, nitric acid and sulfuric acid are mixed, wherein the etching rate of the light absorbing member is adjusted to 10 μm / hour or less,
and so on.
Any one of these may be used, or two or more may be used in combination. When both an alkaline etching solution and an acid etching solution are used, it is preferable that the surface is first roughened or curved with an alkaline etching solution, and then the surface is roughened or curved with an acid etching solution.

As an alkaline etching solution suitable for roughening or curving of the pyramid-shaped tip and side surface, for example,
(A) an etching solution containing 11 to 15% by weight of TMAH and 0.01 to 0.1% by weight of polyethylene glycol;
(B) an etching solution containing 11 to 15% by weight of TMAH and 0.01 to 0.1% by weight of polyethylene glycol mono-4-octylphenyl ether (common name, Triton-X-100);
(C) an etching solution containing 11 to 15% by weight of TMAH and 0.01 to 0.1% by weight of surfactant NC-200,
and so on.

FIG. 2 shows a composition range of an acid etching solution suitable for roughening or curving a pyramidal apex and side surfaces. In the case of a mixed acid of hydrofluoric acid-nitric acid-acetic acid, as shown in FIG. 2 (a), the composition is (HF, HNO ₃ , CH ₃ COOH) = (10, 70, 20), (10, 40 , 50), (20, 30, 50), (30, 30, 40), (25, 45, 30), and (25, 63, 12) are preferably in a region connected by a straight line.
In the case of a mixed acid of hydrofluoric acid-nitric acid-sulfuric acid, as shown in FIG. 2B, the composition is (HF, HNO ₃ , H ₂ SO ₄ ) = ( _2, 38, 60), ( 2, 13, 85), (7, 8, 85), and (20, 20, 60) are preferably in a region connected by a straight line.

[2.3. Diffusion layer forming process]
Next, a first diffusion layer (emitter layer) and, if necessary, a second diffusion layer (BSF layer) are formed on the surface of the light absorbing member (diffusion layer forming step).
As a method of forming the diffusion layer,
(A) A method (solution coating method) in which a layer containing an impurity to be doped is deposited on a surface on which a diffusion layer is to be formed by a method such as coating, and the impurity is diffused by a method such as heat treatment (solution coating method);
(B) a method by ion implantation,
(C) A method of performing a heat treatment while bringing a gas containing impurities into contact with a surface on which a diffusion layer is to be formed (vapor phase diffusion method)
and so on.

Since the diffuse reflection surface is formed on the back surface side of the light absorbing member, it is preferable to use a vapor phase diffusion method when forming a diffusion layer on the back surface (that is, the surface of the diffuse reflection surface). . When the vapor phase diffusion method is used, a diffusion layer having a uniform thickness can be formed on the surface of the diffuse reflection surface having irregularities.
On the other hand, when the diffusion layer is formed on the light receiving surface side surface of the light absorbing member, any of the methods described above may be used.

Specifically, the formation of the diffusion layer using the vapor phase diffusion method is as follows:
(A) Covering a region other than a region for forming the diffusion layer among the front and back surfaces of the light absorbing member with a mask material (for example, silicon dioxide) that does not allow impurity atoms to pass through,
(B) A light absorbing member is inserted into an electric furnace filled with a gas containing impurity atoms, and the impurity atoms are thermally diffused in the light absorbing member in the electric furnace.

There are various methods for filling an electric furnace with a gas containing impurity atoms. For example, when the compound containing impurity atoms is a gas at normal temperature, the gas is fed into the electric furnace as it is or diluted with a carrier gas.
On the other hand, if the compound containing impurity atoms is a liquid at room temperature, but the vapor pressure has a certain level, a carrier gas such as oxygen or nitrogen is allowed to flow through the liquid containing the impurity atoms, and the carrier gas is used as the impurity atoms. Saturated with a liquid vapor containing, and fed into an electric furnace.
Alternatively, when the compound containing impurity atoms is solid at normal temperature but is easily thermally decomposed, the solid containing impurity atoms and the light absorbing member are opposed to each other in the electric furnace and heated in the electric furnace.

When a p-type single crystal silicon substrate is used as the light absorbing member, phosphorus is generally used as the n-type impurity doped in the first diffusion layer (emitter layer). In this case, examples of the phosphorus source that is a gas at normal temperature include phosphine. Examples of the phosphorus source that is liquid at room temperature include phosphorus oxychloride. Regardless of which phosphorus source is used, the carrier gas concentration, temperature, processing time, and the like are controlled so that an appropriate phosphorus concentration distribution is obtained.
In addition, phosphorus glass (phosphosilicate glass) may be formed on the surface of the diffusion layer during the doping of phosphorus. In this case, after removing the phosphorus glass, heat treatment (forcing-in diffusion called “drive-in”) may be performed in an atmosphere of oxygen, nitrogen, or a mixed gas thereof.

When a p-type single crystal silicon substrate is used as the light absorbing member, boron is generally used as the p-type impurity doped into the second diffusion layer (BSF layer). In this case, examples of the boron source that is a gas at normal temperature include diborane. Examples of the boron source that is solid at room temperature include boron nitride. Regardless of which boron source is used, the carrier gas concentration, temperature, processing time, and the like are controlled so that an appropriate boron concentration distribution is obtained.
Further, in the middle of doping with boron, boron glass (borosilicate glass) may be formed on the surface of the diffusion layer. In this case, after removing the boron glass, heat treatment (drive-in) may be performed in an atmosphere of oxygen, nitrogen, or a mixed gas thereof.

  When both the emitter layer and the BSF layer are formed on the surface of the light absorbing member, either the emitter layer or the BSF layer may be formed in any order. However, forming the uneven surface later in the process is not complicated in wafer processing. Therefore, it is preferable to form the diffusion layer on the light receiving surface side first.

[2.4. Electrode / BPF formation process]
Next, a band pass filter and a surface electrode are formed on the light receiving surface side surface of the light absorbing member, and a back surface electrode is formed on the back surface side of the light absorbing member (electrode / BPF forming step).
The front electrode, the back electrode, and the band pass filter can all be formed using a known thin film lamination technique. In this case, the order of forming the thin film on the light receiving surface side and the thin film on the back surface side is not particularly limited.

  Moreover, when forming a band pass filter and a surface electrode in the light-receiving surface side, the formation order is not specifically limited, either. That is, a band pass filter may be formed first, and a surface electrode may be formed around it. Alternatively, the surface electrode may be formed first, the opening may be formed by etching in a portion where the band pass filter is to be formed, and the band pass filter may be formed in the opening.

[3. Action]
[3.1. Light confinement effect]
FIG. 3 shows a schematic cross-sectional view of a non-reflective surface shape (texture) provided on the light-receiving surface side of a conventional solar cell. As shown in FIG. 3A, in the conventional solar cell, regular irregularities are formed on the surface of the semiconductor substrate on the light receiving surface side. When light is incident on the side surface of the first convex portion, a large part of the light enters the semiconductor substrate (first intrusion light), and part of the light is reflected on the side surface of the first convex portion (first reflected light). If the height and size of the irregularities formed on the semiconductor substrate are evenly distributed, the first reflected light hits the second convex portion and generates second intrusion light and second reflected light. Each time light is incident on the convex portion, a large portion of the light enters the semiconductor substrate, so that the intensity of the reflected light greatly decreases as the incidence on the convex portion is repeated.

  On the other hand, as shown in FIG. 3B, when the height of the convex portion is not uniform, the first reflected light may not strike the second convex portion. In this case, since the first reflected light does not enter the semiconductor substrate, it does not contribute to power generation. For the above reasons, in the conventional solar cell, it is necessary to form irregularities as regular as possible on the surface on the light receiving surface side.

  In FIG. 4, the cross-sectional schematic diagram of the diffuse reflection surface provided in the back surface side of the photoelectric conversion element is shown. The photoelectric conversion element according to the present invention is characterized in that a band-pass filter (BPF) and a surface electrode are formed on the light receiving surface side of the light absorbing member, and a diffuse reflection surface and a back electrode are formed on the back surface side.

  BPF has been conventionally used to extract light of a specific wavelength from white light. When this BPF is provided on the light receiving surface of the photoelectric conversion element and monochromatic light is incident on the light receiving surface, the BPF reflects the reflected light from the back surface again into the light absorbing member without blocking the incident light. Thereby, even if the light absorbing member is thinned, a high light absorption rate is secured. Further, the material can be reduced and the tact time can be shortened, so that the cost can be reduced.

  Further, when the periphery of the BPF is covered with a surface electrode of a predetermined size and a diffuse reflection surface and a back electrode are provided so as to face the BPF and the surface electrode, the incident light repeats multiple reflections between them. The inside of the light absorbing member is diffused in a direction parallel to the light receiving surface. Therefore, the absorption rate of incident light by the light absorbing member is improved.

  However, when a diffuse reflection surface is formed on the back surface side of the light absorbing member, the inclination angle and the size of the unevenness are uniform and the unevenness is regularly arranged, as shown in FIG. Reflects in a certain direction. Therefore, light does not diffuse and propagate widely in the light absorbing member. In particular, when the inclination angle of the unevenness is close to 45 ° (for example, in the range of 35 ° to 55 °), the light reflected several times on the uneven surface reenters the BPF at an angle smaller than the critical angle of the BPF. Therefore, incident light leaks outside the photoelectric conversion element.

  On the other hand, as shown in FIG. 4B, when the uneven surface has various sizes and inclination angles, the light incident on the light absorbing member from the light receiving surface is reflected by the uneven surface in various directions. Therefore, light diffuses and propagates widely in the light absorbing member. As a result, light is sufficiently absorbed by the light absorbing member, which can sufficiently contribute to the conversion efficiency of the photoelectric conversion element. In particular, when the shape of the diffuse reflection surface satisfies the above-described expression (1), the absorption rate of incident light by the light absorbing member is further improved.

[3.2. Diffuse reflective surface]
As a method for forming irregularities on the surface of a light absorbing member (semiconductor substrate), for example,
(A) a first method using an alkaline etching solution;
(B) a second method using an acid etching solution;
(C) a third method combining mechanical or chemical damage formation and acid etching;
Etc. are known.

Of these, the first method tends to form regular irregularities.
Since the second method is isotropic etching, it is difficult to form an uneven surface unless there is a difference in the microcrystalline structure of the semiconductor substrate. Depending on the composition of the acid etching solution, the mask material may be dissolved or peeled off during etching.
Furthermore, the third method not only complicates the manufacturing process, but sometimes damages the crystal structure of the light absorbing member microscopically. For this reason, the conversion efficiency of the photoelectric conversion element is reduced due to an increase in carrier recombination or the like.

  On the other hand, first, a pyramid having a random size is formed on the back surface of the light absorbing member by using the first etching solution, and then the tip and side surfaces of the pyramid are roughened by using the second etching solution. Alternatively, using a curved surface method, an ideal diffuse reflection surface can be formed relatively easily. Further, since this method has a high degree of freedom in selecting a mask material, a diffuse reflection surface can be selectively formed on the back surface of the light absorbing member. Furthermore, since no micro damage is given to the crystal structure of the light absorbing member, carrier recombination is also suppressed.

[3.3. Diffusion layer]
FIG. 5 shows an energy band diagram of a solar cell in which the first diffusion layer (1) exists in the vicinity of the light receiving surface. In many conventional solar cells, a first diffusion layer (emitter layer) (1) is formed on the light-receiving surface side of a semiconductor substrate. The reason is that the density of carriers generated when the semiconductor substrate absorbs light is large in the portion close to the light receiving surface. Therefore, by providing the pn junction surface (3) in the portion close to the light receiving surface, This is because carriers can be taken out from the electrodes. The pn junction surface (3) serves as a driving force for separating and accelerating carriers generated by light in the respective directions of the two electrodes (+ electrode and −electrode) of the solar cell.
Further, in the solar cell in which the pn junction surface (3) is provided near the light receiving surface, the minority carriers (electrons in the case of FIG. 5) to be diffused to the light receiving surface are prevented from diffusing and moving to the back surface. Therefore, the second diffusion layer (BSF layer) (2) is often provided on the back surface of the semiconductor substrate.

As a method for forming such a diffusion layer, as described above, a solution coating method, an ion implantation method, a vapor phase diffusion method, and the like are known. Among these, even if the solution application method uniformly applies the solution to the concavo-convex surface, the solution eventually accumulates in the concave portion and does not remain in the convex portion with a sufficient thickness. Therefore, in the solution coating method, it is difficult to dope the uneven surface with a uniform concentration of impurities.
In the ion implantation method, the penetration depth of ions is substantially determined by the mass and energy of implanted ions and the material of the substrate. For this reason, when the concavo-convex surface is doped with an impurity by the ion implantation method, the diffusion layer becomes thick on the flat surface and thin on the inclined surface.
Therefore, when a diffusion layer is formed on the uneven surface, it is preferable to use a vapor phase diffusion method.

When the light absorbing member is made of p-type single crystal silicon, an emitter layer to which a vapor phase diffusion method can be applied is preferably formed on the back surface side of the light absorbing member. There is an example in which an emitter layer is provided on the back surface of a conventional solar cell (see Patent Document 5). However, in the solar cell described in this document, the BSF layer is also formed on the back surface. By forming both the emitter layer and the BSF layer on the back surface, it is not necessary to provide an electrode (generally a metal electrode, which reflects light outside the solar cell) on the light receiving surface side, so that a decrease in efficiency is suppressed to some extent. Can do. However, since the pn junction surface cannot be formed over the entire light receiving width, a part of efficiency is sacrificed.
In the present invention, since the pn junction surface can be formed over the entire light receiving width, the conversion efficiency is not sacrificed unlike the conventional solar cell.

[3.4. Light diffusion part]
Light incident from the bandpass filter is reflected by the diffuse reflection surface and the back electrode on the back surface side. The reflected light is re-reflected by the band-pass filter and the surface electrode on the light receiving surface side. While repeating such multiple reflection, light diffuses in a direction parallel to the light receiving surface. In the present invention, in order to increase the light absorptance, the light diffusing portion has a structure suitable for multiple reflection, and at the same time, the light diffusing portion is thinned. Therefore, even when the first diffusion layer is formed on the back surface side, the incident light reaches the first diffusion layer with almost no attenuation. This can be explained by the following calculation.

In the photoelectric conversion element according to the present invention, when the wavelength λ _{0 of} incident light is 1064 nm, the extinction coefficient k of single crystal silicon is k = 6.7 × 10 ⁻⁵ (see Reference 1). Therefore, the penetration depth into the single crystal silicon having a wavelength of 1064 nm is calculated to be about 1.3 mm. In the present invention, the thickness of the light diffusing portion is set to 1/200 to 1/10 of the penetration depth of incident light, so the preferred thickness is 6.5 to 130 μm.

  In single crystal silicon, the thickness of the light diffusion portion considered to be most preferable practically is 20 to 80 μm. When the incident light reaches the bottom surface of the single crystal silicon having a thickness of 20, 50, and 80 μm, the energy decrease rate of the incident light is only 1.6, 3.9, and 6.1%, respectively. Therefore, the distribution of incident light energy in the thickness direction may be considered to be substantially uniform. Therefore, the density of carriers generated when incident light is absorbed by single crystal silicon may be considered to be substantially uniform in the cross section in the thickness direction. Therefore, even if the first diffusion layer is provided on the back surface of the single crystal silicon, the conversion efficiency of the photoelectric conversion element does not decrease so much that there is a practical impediment.

[3.5. Multilayer electrode]
In the present invention, the electrode is required not only to function as a current collector for extracting carriers but also to function as a reflective film that reflects incident light. Among the metal materials, silver has the highest reflectance. When the light absorbing member is p-type single crystal silicon, a Schottky junction is not formed even when silver and the first diffusion layer (emitter layer) are brought into contact with each other. Therefore, silver can be used for the electrode in contact with the first diffusion layer. However, as shown in FIG. 6, when silver is directly connected to the second diffusion layer, there is a problem that a Schottky junction is formed, which is inconvenient as an electrode.

  In order to solve this problem, it is also considered to use a metal that does not form a Schottky junction with the light absorbing member (for example, Al in the case of p-type single crystal silicon) as the electrode material in contact with the second diffusion layer. It is done. However, even if there is a metal that can avoid the formation of a Schottky junction, the metal does not always have sufficient reflectivity for incident light. A slight decrease in reflectance causes a large decrease in conversion efficiency.

  In contrast, the electrode is made of a first layer made of a first conductive material capable of forming an ohmic junction with the light absorbing member, and a second conductive material having a higher reflectance of the incident light than the first conductive material. With a stacked structure with the second layer, formation of a Schottky junction and a decrease in reflectance can be suppressed at the same time. In addition, a highly reflective material can always be selected as the second layer material regardless of the composition of the light-absorbing material, increasing design freedom.

(Example 1: Etching using tetramethylammonium hydroxide aqueous solution)
[1. Preparation of sample]
Etching of p-type single crystal silicon was performed using a tetramethylammonium hydroxide (TMAH) aqueous solution. The TMAH concentration in the TMAH aqueous solution was 10 to 25% by weight. The liquid temperature during etching was 88 ° C. to 90 ° C.

[2. result]
FIG. 7 shows the relationship between the etching rate and the concentration when p-type single crystal silicon is etched with a TMAH aqueous solution (liquid temperature 90 ° C.). FIG. 8 shows a scanning electron microscope (SEM) photograph of the etched surface when etched using an aqueous TMAH solution at an etching rate of 50 to 60 μm / h for 3 hours. 7 and 8 show the following.

(1) The etching rate can be adjusted by changing the TMAH concentration in the TMAH aqueous solution. When the TMAH concentration in the TMAH aqueous solution is optimized, an etching rate of 50 μm / h or more can be obtained at a temperature of about 90 ° C.
(2) When etching was performed using a TMAH aqueous solution at an etching rate of 50 μm / h or more, a pyramid structure may be partially observed (FIG. 8A). However, the etched surface was generally smooth (FIG. 8 (b)). Therefore, a TMAH aqueous solution having an etching rate of 50 μm / h or more can be used as an etching solution for forming a thin plate portion.

(Example 2: Etching using acid etching solution)
[1. Preparation of sample]
Etching of p-type single crystal silicon was performed using a mixed solution of hydrofluoric acid-nitric acid-sulfuric acid or a mixed liquid of hydrofluoric acid-nitric acid-acetic acid. The etching time was 90 seconds. The composition ratio of the acid in the acid etching solution is as follows.
(A) HF: HNO ₃ : H ₂ SO ₄ = 4: 16: 80 (Sample No. 1, FIG. 9 (a)).
(B) HF: HNO ₃ : H ₂ SO ₄ = 10: 15: 75 (Sample No. 2, FIG. 9 (b)).
(C) HF: HNO ₃ : CH ₃ COOH = 18: 36: 46 (Sample No. 3, FIG. 9 (c)).

[2. result]
9 (a) to 9 (c), sample No. 1 to Sample No. 3 shows an SEM photograph of the Si surface after the etching of No. 3; FIG. 9 shows the following.
(1) Although acid etching tends to be isotropic etching, random irregularities that can be used as a diffuse reflection surface can be formed by optimizing the liquid composition.
(2) When etched with the etching solution (c), the diameter of the recess formed on the surface was 5 μm or more. In the case of the etching solution (b), the diameter of the recess was 2 to 5 μm. Further, in the case of the etching solution (a), the diameter of the recess was 1 to 2 μm. In other words, the diameter of the recess became closer to the wavelength of the incident light in the order of the etching solution (c) → (b) → (a). Since the light is reflected in various directions on the uneven surface close to the wavelength of the incident light, the etching solution (a) is most preferable.

(Example 3: Formation of diffuse reflection surface by two-step etching)
[1. Preparation of sample]
First, p-type single crystal silicon was etched (thinned) using a TMAH aqueous solution. The TMAH concentration in the TMAH aqueous solution was 15% by weight. The liquid temperature during etching was 90 ° C.
Next, etching (pyramid formation) of the thinned portion was performed using the first etching solution. As the first etching solution, a TMAH aqueous solution having a TMAH concentration of 11% by weight was used. The etching time was 1 hour. Furthermore, the liquid temperature at the time of etching was 90 ° C. (Sample No. 11, FIG. 10A).

Furthermore, the etching (roughening, curved surface) of the point part and side surface of the pyramid was performed using the second etching solution. Etching conditions are as follows.
(A) Etching (roughening) for 1 hour with a 15% by weight TMAH aqueous solution (liquid temperature: 90 ° C.) containing 0.05% by weight polyethylene glycol mono-4-octylphenyl ether (sample No. 12, FIG. 10 (b)).
(B) Etching (roughening) for 1 hour with a 15% by weight TMAH aqueous solution (liquid temperature: 90 ° C.) containing 0.05% by weight polyethylene glycol mono-4-octylphenyl ether, followed by HF: HNO ₃ : CH ₃ COOH = 20: 40: 40 etching with acid etching solution (liquid temperature: 23 ° C.) for 2 minutes (curved surface) (Sample No. 13, FIG. 10 (c)).
(C) Etching (roughening) for 90 seconds with an acid etching solution (liquid temperature: 23 ° C.) of HF: HNO ₃ : H ₂ SO ₃ = 4: 16: 80 (Sample No. 14, FIG. 10 (d)) .

[2. result]
10 (a) to 10 (d), sample No. 11 to Sample No. 14 SEM photographs are shown. FIG. 10 shows the following.
(1) When etched with an 11 wt% TMAH aqueous solution, random-sized pyramids can be formed on the entire Si surface (FIG. 10A).
(2) The surface of the pyramid can be roughened or curved when Si on which a pyramid having a random size is formed is further etched with an alkaline etching solution and / or an acid etching solution having a predetermined composition (see FIG. 10 (b) to FIG. 10 (d)).

(Example 4: Measurement of reflection intensity)
[1. Preparation of sample]
Using the etching method described in Example 2 and Example 3, the following reflection intensity measurement samples were prepared.
(1) Sample No. 21: Thin plate only.
(2) Sample No. 22: Thinning + pyramid formation (11 wt% TMAH).
(3) Sample No. 23: Thinning + acid etching (HF: HNO ₃ : H ₂ SO ₄ = 10: 15: 75).
(4) Sample No. 24: Thinning + acid etching (HF: NHO ₃ : H ₂ SO ₄ = 4: 16: 80).
(5) Sample No. 25: Thin plate + pyramid formation (11 wt% TMAH) + curved surface (15 wt% TMAH (Triton))
(6) Sample No. 26: Thinning + pyramid formation (11 wt% TMAH) + roughening (HF: NHO ₃ : H ₂ SO ₄ = 4: 16: 80).

[2. result]
When light with a wavelength of 1064 nm is incident on the thinned silicon substrate (sample No. 21) and the silicon substrate (samples No. 22 to No. 26) with unevenness formed on the surface in FIG. Shows the dependence of the normalized reflection intensity (I / I ₀ ) on the reflection angle. FIG. 11B shows the relationship between the reflection angle θ and (I / I _o ) · sin θ. In FIG. 11, “cos (θ)” represents complete diffuse reflection. FIG. 11 shows the following.
(1) In the case of only thinning (sample No. 21), the back surface was partially formed with a pyramid structure, but was flat in most regions. Therefore, the incident light was reflected almost as it was in the incident direction. That is, the full width at half maximum of (I / I _o ) · sin θ was 6.2 °.

(2) After thinning, when etched with an 11 wt% TMAH aqueous solution (Sample No. 22), a pyramid shape was formed on the entire etching surface. Since the side surface of the pyramid shape is a flat surface having a certain inclination, reflection in a specific angular direction becomes remarkable. The full width at half maximum of (I / I _o ) · sin θ at this time was 6.4 °.
(3) When acid etching is performed after thinning (Sample No. 23, Sample No. 24), unevenness as shown in FIGS. 9 (a) and 9 (b) is formed on the silicon substrate surface. It was. Therefore, the reflection angle dependency distributed widely in the angle range up to 90 ° was obtained. The full widths at half maximum of (I / I _o ) · sin θ at this time were 62.2 ° and 60.6 °, respectively.
On the other hand, sample No. 25, the value of (I / I ₀ ) · sin θ was widely distributed up to θ ≦ 80 °, but the full width at half maximum was 14.3 °.

(4) When the acid etching was further performed after forming the pyramid shape (Sample No. 26), the side surface of the pyramid was roughened as shown in FIG. Therefore, the reflection intensity increased remarkably over an angle range of 30 ° to 90 °.
Further, this reflection intensity is the same as that of Sample No. Compared with 24, the reflection intensity increased in the angle range of 50 ° to 90 °. As a result, the full width at half maximum of (I / I _o ) · sin θ reached 72.0 °. Therefore, it has been found that roughening by acid etching after forming the pyramid shape gives more favorable results.
(5) A wide distribution of reflection angles is important for optical confinement. The angle (polar angle θ) distribution of the reflection intensity is approximately isotropic in the azimuth direction, and is proportional to a value obtained by multiplying the measured reflection intensity by sin (θ). Sample No. In 22, strong reflection occurs in the direction of about 22 ° from the incident direction, but the distribution width is narrow.
In contrast, sample no. In 23-26 (except for sample No. 25), a full width at half maximum of 50 ° or more was obtained.
(6) Assuming that the angular distribution of the reflection intensity is cos ⁿ (θ), a simulation of the optical absorptance was performed when 1064 nm light was incident on a 50 μm thick Si element with a surface bandpass filter. The result is FIG. In this case, if n ≦ 2, the half width at half maximum is 50 ° or more, and if n ≦ 8, the full width at half maximum is 30 ° or more, and the light absorptance is 0.83 or more and 0.80 or more, respectively.

(Example 5: Multilayer electrode)
[1. Preparation of sample]
A second diffusion layer was formed on the p-type single crystal silicon substrate. Next, a laminated electrode having an aluminum thickness of 1.8 nm and a silver thickness of 500 nm was formed on the second diffusion layer. Further, the substrate was subjected to sintering treatment under the conditions of temperature: 400 ° C. (sample No. 31) or 420 ° C. (sample No. 33) and treatment time: 15 minutes.
Similarly, a laminated electrode having an aluminum thickness of 2.7 nm and a silver thickness of 500 nm was formed on the second diffusion layer. Further, the substrate was sintered under conditions of a temperature of 400 ° C. and a processing time of 15 minutes (Sample No. 32).

[2. result]
[2.1. Current-voltage characteristics]
FIG. 13A shows current-voltage characteristics when the second diffusion layer is formed on the p-type single crystal silicon substrate and the silver / aluminum laminated electrode is formed on the second diffusion layer. In FIG. 13B and FIG. 13C, a p-type formed with a silver / aluminum laminated electrode having an aluminum thickness of 1.8 nm (sample No. 31) or 2.7 nm (sample No. 32), respectively. The transmission electron micrograph of the cross section of a single crystal silicon substrate is shown. FIG. 13 shows the following.
(1) Even if the aluminum thickness is 2.7 nm or less, ohmic characteristics can be obtained if the sintering process conditions are 400 ° C. × 15 minutes (Sample Nos. 31 and 32).
(2) When the sintering temperature is 420 ° C., silver atoms reach the surface of the second diffusion layer by silver-aluminum interdiffusion. Therefore, a Schottky barrier begins to be formed, and rectification characteristics begin to appear in the current-voltage curve (Sample No. 33). Therefore, the sintering temperature is preferably determined within a range not exceeding 420 ° C.

[2.2. Light reflectivity]
FIG. 14 shows the relationship between the thickness of the aluminum layer of the silver / aluminum multilayer electrode and the reflectance when light having a wavelength of 1064 nm travels from the second diffusion layer toward the silver / aluminum multilayer electrode. FIG. 14 shows that the reflectance exceeds 90% when the thickness of the aluminum layer is 3 nm or less. Therefore, if the aluminum layer has a thickness of 3 nm or less and is subjected to a sintering process for 15 minutes within a range not exceeding 420 ° C., a Schottky junction is not formed and a high light confinement effect is exhibited. An electrode structure can be formed.
Note that the electrode in contact with the first diffusion layer (n-type) of the p-type single crystal silicon substrate has no problem as a single silver. Since silver is a highly reflective material having a reflectance exceeding 95%, light can be sufficiently confined in the thin plate portion of the substrate.

(Example 6: Size of light diffusion portion)
Consider a case where a light beam having a wavelength of 1064 nm and a diameter of 50 μm or 100 μm is perpendicularly incident on a Si photoelectric conversion element etched to a thickness of 20 μm or 50 μm.

[1. Diffusion distance]
When incident light (not a light beam) was multiple-reflected on the diffuse reflection surface and the surface with the bandpass filter, the distance propagating in the lateral direction was examined by the ray tracing method. The diffuse reflection phenomenon on the back surface was handled by the Monte Carlo method.
FIG. 15 shows the relationship between the distance (propagation distance) from the incident position of light and the light intensity at that point. This was fitted with an exponential function to obtain a diffusion distance (a distance at which the light intensity becomes 1 / e). From FIG. 15, the diffusion distances of Si having a thickness of 20 μm and 50 μm were determined to be 86 μm and 130 μm, respectively.

[2. Incident beam intensity distribution]
The distribution of the incident light intensity when the light beam entered the element was determined. FIG. 16 shows the relationship between the distance from the center of the light beam having a diameter of 50 μm (FIG. 16A) or 100 μm in diameter (FIG. 16B), the light intensity at that point, and the light intensity at that point. Note that the beam diameter = 2 × (beam radius). From FIG. 16, it can be seen that light having an intensity of 0.05 times or more of the incident light intensity reaches from the edge of the incident light beam to a range of about 2 to 3 times the diffusion distance of the light.

[3. Si wafer thickness]
The relationship between the thickness of the Si wafer and the conversion efficiency of the element was calculated using simulation software PC1D. At this time, the incident light intensity was assumed to be 1 kW / cm ² and uniform in the in-plane direction.
FIG. 17 shows the relationship between the thickness of the Si wafer and the conversion efficiency. FIG. 17 shows the following.
(A) When the thickness of the Si wafer is 100 μm or less, the conversion efficiency exceeds 40%.
(B) When the thickness of the Si wafer is 50 μm or less, the conversion efficiency exceeds 55%.
(C) When the thickness of the Si wafer is about 20 μm, the conversion efficiency is maximized.

[4. Radius of light diffusion part]
As shown in FIG. 16, the conversion efficiency η of the element in the case where the light intensity is distributed and the light exceeds the range of the back surface etching was calculated using an approximation such as the following equation.
η = (η _a × I _a + η _b × I _b ) / I _total
However,
η _a is the conversion efficiency of the etched part,
I _a is the integrated light intensity of the etched part,
η _b is the conversion efficiency of the non-etched part,
I _b is the integrated light intensity of the non-etched part,
I _total is the total of light intensity in all areas.
As η _a , 61.1% (thickness: 20 μm) or 56.0% (thickness: 50 μm) was used. Further, 25.7% (thickness: 200 μm) was used as η _b .

  FIG. 18 shows the relationship between the radius of the etched portion and the conversion efficiency when a light beam having a diameter of 50 μm (FIG. 18A) or a diameter of 100 μm (FIG. 18B) is incident on Si whose back surface is etched. Show. As can be seen from FIG. 18, the radius of the etched portion (light diffusion portion) needs to be 2 to 3 times the beam radius + the diffusion distance in order to obtain high conversion efficiency.

  Although the embodiments of the present invention have been described in detail above, the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the gist of the present invention.

  The photoelectric conversion element according to the present invention can be used for solar cells, photoconductive cells, photodiodes, phototransistors and the like.

10 Photoelectric Conversion Element 12 Light Absorbing Member 12a Second Diffusion Layer (Back Surface Field Layer)
12b Diffuse reflecting surface 12c First diffusion layer (emitter layer)
14 Band pass filter 16 Front electrode 18 Back electrode

Claims (9)

A photoelectric conversion element having the following configuration.
(1) The photoelectric conversion element includes a light-absorbing member made of a light-absorbing material capable of absorbing monochromatic light having a wavelength λ ₀ and generating carriers, and substantially parallel rays made of the monochromatic light as incident light. Is used.
(2) The photoelectric conversion element is
The light absorbing member;
A band pass filter formed on the light receiving surface side surface of the light absorbing member;
A surface electrode formed around the bandpass filter;
A surface on the back side of the light absorbing member, a diffuse reflection surface formed at a position facing the band pass filter and the surface electrode;
A back electrode formed so as to cover all or part of the surface of the diffuse reflection surface.
(3) The band-pass filter has a function of selectively transmitting at least light having the wavelength λ ₀ .
(4) The diffuse reflection surface has a function of diffusing the incident light in a direction parallel to the light receiving surface by multiple reflection.
(5) The front electrode and the back electrode each have a function as a current collector for extracting the carriers and a function as a reflective film for reflecting the incident light. The photoelectric conversion element according to claim 1, further comprising the following configuration.
(6) The thickness of the light diffusion part of the light absorbing member is 1/200 or more and 1/10 or less of the penetration depth of the incident light into the light absorbing material.
Here, the “light diffusing portion” means that when the light absorbing member is viewed from the normal direction of the light receiving surface, the projection surface of the bandpass filter and the surface electrode and the projection surface of the diffuse reflection surface are An overlapping part.
The “depth of penetration of incident light into the light absorbing material” is a depth at which the intensity of the incident light becomes 1 / e (e is the number of Napier) when the incident light is incident on the light absorbing material. Say.
(7) The shortest distance from the edge of the bandpass filter to the edge of the light diffusing unit is at least twice the diffusion distance.
Here, the “diffusion distance” means that the band-pass filter and the diffusive reflection surface are formed on the front and back surfaces of the light-absorbing material having a thickness corresponding to the thickness of the light diffusion portion, respectively. When the incident light is diffused in a direction parallel to the light receiving surface while multiple reflection of the incident light is performed between a band pass filter and the diffuse reflection surface, the intensity of the incident light is 1 / e (e is , Napier number). The photoelectric conversion element according to claim 1, wherein the diffuse reflection surface has a reflection spectrum of the incident light satisfying a relationship expressed by the following expression (1).
Full width at half maximum of (I / I ₀ ) · sinθ ≧ 30 ° (1)
Where I / I ₀ is the normalized reflection intensity, and θ is the reflection angle measured from the incident direction. The diffuse reflection surface is
Using the first etching solution, a random pyramid is formed on the back surface of the light absorbing member,
The photoelectric conversion element according to any one of claims 1 to 3, wherein the photoelectric conversion element is obtained by roughening or curving the tip and side surfaces of the pyramid using a second etching solution.   The photoelectric conversion element according to claim 4, wherein the first etching solution is an alkaline etching solution that contains neither alkali metal ions nor alkaline earth metal ions. The second etchant is
(A) An alkaline etching solution containing neither alkali metal ions nor alkaline earth metal ions, wherein an additive having a chain molecular structure in which one end is hydrophobic and the other end is hydrophilic is added to the light. What adjusted the etching rate of the absorbing member to 15 μm / hour or less,
(B) An acid etching solution in which hydrofluoric acid, nitric acid and acetic acid are mixed, wherein the etching rate of the light absorbing member is adjusted to 10 μm / hour or less, and / or
(C) An acid etching solution in which hydrofluoric acid, nitric acid and sulfuric acid are mixed, wherein the etching rate of the light absorbing member is adjusted to 10 μm / hour or less,
The photoelectric conversion element according to claim 4 or 5. The photoelectric conversion element according to any one of claims 1 to 6, further comprising the following configuration.
(8) At least one of the front electrode and the back electrode is
A first layer made of a first conductive material formed on a surface of the light absorbing member and capable of forming an ohmic junction with the light absorbing member;
A laminated structure with a second layer made of a second conductive material formed on the surface of the first layer and having a higher reflectance of the incident light than the first conductive material;
The thickness of the first layer is not less than three times the penetration depth of carriers into the first conductive material, and is not more than the penetration depth of the incident light into the first conductive material.
Here, “the penetration depth of the carrier into the first conductive material” is assumed that the carrier (A) in the light absorbing material and the carrier (B) in the first conductive material can be distinguished, When the concentration distribution in the first conductive material of the carrier (A) in the light absorbing material is calculated using Poisson's equation, the concentration of the carrier (A) in the first conductive material is The depth which becomes 1 / e (e is the Napier number) of the concentration of the carrier (A) in the absorbent material.
“Invasion depth of incident light into the first conductive material” means that when the incident light is incident on the first conductive material, the intensity of the incident light is 1 / e (e is the number of Napier). Say depth.   The photoelectric conversion element according to claim 1, wherein each of the front electrode and the back electrode has a reflectance of 90% or more for the incident light. The photoelectric conversion element according to any one of claims 1 to 8, further comprising the following configuration.
(9) A first diffusion layer (emitter layer) having a conductivity type different from that of the light absorbing member is formed on the back side of the light absorbing member.

Images