JP2005340852A - Semiconductor device and electronic apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a manufacturing method of a semiconductor device, and a semiconductor device manufactured by using the manufacturing method, in which a laser crystallization method is used that is capable of preventing the formation of grain boundaries in a TFT channel formation region, and capable of preventing conspicuous drops in TFT mobility, reduction in an on-current and increases in an off-current, that are caused by the grain boundaries. <P>SOLUTION: Depressions and projections with stripe shapes or rectangular shapes are formed. Continuous oscillation laser light is then made to irradiate a semiconductor film formed on the insulating film along the depressions and projections with the stripe shapes of the insulating film, or along the longitudinal axis direction or the minor axis direction of the rectangular shape. Although it is most preferable to use continuous oscillation laser light at this point, pulse oscillation laser light may also be used. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a semiconductor device formed using a semiconductor film having a crystal structure and a manufacturing method thereof, and an island-shaped semiconductor region including a channel formation region is formed using a crystalline semiconductor film formed over an insulating surface. The present invention relates to a field effect transistor, in particular, a semiconductor device including a thin film transistor and a manufacturing method thereof.

  In recent years, a technology for forming a TFT on a substrate has greatly advanced, and application development to an active matrix semiconductor display device has been advanced. In particular, a TFT using a polycrystalline semiconductor film has higher field effect mobility (also referred to as mobility) than a TFT using a conventional amorphous semiconductor film, and thus can operate at high speed. For this reason, it is possible to perform control of a pixel, which has been conventionally performed by a drive circuit provided outside the substrate, with a drive circuit formed on the same substrate as the pixel.

  By the way, as a substrate used for a semiconductor device, a glass substrate is considered promising rather than a single crystal silicon substrate in terms of cost. A technique is known in which an amorphous silicon film is formed on an insulating substrate made of glass or the like and crystallized by laser processing. A glass substrate is inferior in heat resistance and easily deforms by heat. Therefore, in the case where a polysilicon TFT is formed on a glass substrate, using laser annealing for crystallization of the semiconductor film is very effective for avoiding thermal deformation of the glass substrate. Thin film transistors (hereinafter referred to as TFTs) manufactured using the crystalline silicon film are applied to, for example, liquid crystal display devices.

  The characteristics of laser annealing are that the processing time can be significantly shortened compared to annealing methods that use radiation heating or conduction heating, and that the substrate or semiconductor film is selectively and locally heated to cause almost no thermal damage to the substrate. Is not given.

  The laser annealing method referred to here is a technique for recrystallizing a damaged layer formed on a semiconductor substrate or a semiconductor film, a technique for crystallizing a semiconductor film formed on a substrate, or a semiconductor film having a crystal structure. It refers to a technique for improving the crystallinity of (crystalline semiconductor film). Moreover, the technique applied to planarization and surface modification of a semiconductor substrate or a semiconductor film is also included. The laser oscillation device to be applied is a gas laser oscillation device typified by an excimer laser or a solid-state laser oscillation device typified by a YAG laser. A semiconductor surface layer is irradiated with laser light to tens of nanoseconds to tens of microseconds. It is known to be crystallized by heating for a very short time.

  An example of crystallization of an amorphous semiconductor film by laser light irradiation is as follows. As disclosed in Patent Document 1 below, the amorphous semiconductor film is amorphous by high-speed scanning with a laser beam scanning speed of beam spot diameter × 5000 / second or more. A semiconductor film that is polycrystallized without reaching a complete molten state, or a semiconductor region formed in the following Patent Document 2 is irradiated with an extended laser beam to form a substantially single crystal. A technique for forming a region is disclosed. Alternatively, a method is known in which a beam is processed and irradiated linearly by an optical system as in a laser processing apparatus disclosed in Patent Document 3 below.

Furthermore, as disclosed in the following Patent Document 4, a solid-state laser oscillation device such as an Nd: YVO ₄ laser is used to irradiate an amorphous semiconductor film with a laser beam that is the second harmonic, compared to the conventional case. A technique for forming a TFT by forming a crystalline semiconductor film having a large crystal grain size is disclosed.
JP-A-62-104117 (page 92) US Pat. No. 4,330,363 (FIG. 4) JP-A-8-195357 (page 3-4, FIG. 1-5) JP 2001-144027 A (page 4)

Attempts to form a single crystal semiconductor film on an insulating surface have been made for a long time, and a technique called graphoepitaxy has been devised as a more aggressive attempt. Graphoepitaxy forms a step on the surface of a quartz substrate, forms an amorphous semiconductor film or a polycrystalline semiconductor film on it, and then heats it with a laser beam or heater to form a step shape formed on the quartz substrate. This is a technique of forming an epitaxial growth layer with nuclei as a nucleus. This technique is disclosed in Non-Patent Document 1, for example.
J. Vac. Sci. Technol., "Grapho-epitaxy of silicon on fused silica using surface micropatterns and laser crystallization", 16 (6), 1979, pp1640-1643.

For example, Non-Patent Document 2 also discloses a semiconductor film crystallization technique called graphoepitaxy. This was an attempt to epitaxially grow a semiconductor film by induction of a surface relief grating on the surface of an artificially produced amorphous substrate. In Non-Patent Document 3, the technique of graphoepitaxy is that a step is provided on the surface of the insulating film, and the semiconductor film formed on the insulating film is subjected to processing such as heating or irradiation with laser light. It is disclosed that the crystal of the semiconductor film is epitaxially grown.
MW Geis, et al., "CRYSTALLINE SILICON ON INSULATORS BY GRAPHOEPITAXY" Technical Digest of International Electron Devices Meeting, 1979, pp.210.

Lasers are roughly classified into two types, pulse oscillation and continuous oscillation, depending on the oscillation method. Since the pulsed laser has a relatively high output energy, the mass productivity can be improved by setting the size of the laser beam to several cm ² or more. In particular, when the shape of the laser beam is processed using an optical system so as to be a linear shape having a length of 10 cm or more, the substrate can be efficiently irradiated with laser light, and mass productivity can be further improved. For this reason, it has become the mainstream to use a pulsed laser for crystallization of the semiconductor film.

  However, in recent years, it has been found that the crystal grain size formed in a semiconductor film is larger when a continuous wave laser is used than when a pulsed laser is used for crystallization of a semiconductor film. As the crystal grain size in the semiconductor film increases, the mobility of TFTs formed using the semiconductor film increases. For this reason, continuous wave lasers are starting to attract attention.

  A crystalline semiconductor film manufactured using a laser annealing method roughly divided into pulse oscillation and continuous oscillation is generally formed by aggregating a plurality of crystal grains. The positions and sizes of the crystal grains are random, and it is difficult to form a crystalline semiconductor film by specifying the positions and sizes of the crystal grains. Therefore, there may be a crystal grain interface (grain boundary) in the active layer formed by patterning the crystalline semiconductor film into an island shape.

  Unlike crystal grains, there are innumerable recombination centers and trap centers due to amorphous structures and crystal defects at grain boundaries. It is known that when carriers are trapped in this trapping center, the grain boundary potential increases and becomes a barrier against the carriers, so that the current transport characteristics of the carriers decrease. Therefore, if there is a grain boundary in the active layer of the TFT, particularly in the channel formation region, the mobility of the TFT is significantly reduced, the on-current is reduced, and the current flows at the grain boundary, so that the off-current increases. However, it has a significant effect on the TFT characteristics. In addition, in a plurality of TFTs manufactured on the assumption that the same characteristics can be obtained, the characteristics vary depending on the presence or absence of grain boundaries in the active layer.

  The reason for the random position and size of the crystal grains obtained when the semiconductor film is irradiated with laser light is as follows. A certain amount of time is required until solid-phase nucleation occurs in the liquid semiconductor film completely melted by the laser light irradiation. As time passes, innumerable crystal nuclei are generated in the complete melting region, and crystals grow from the crystal nuclei. Since the positions where the crystal nuclei are generated are random, the crystal nuclei are unevenly distributed. Since crystal growth ends when the crystal grains collide with each other, the position and size of the crystal grains are random.

  Therefore, it is ideal to form the channel formation region having a significant influence on the characteristics of the TFT with a single crystal grain by eliminating the influence of the grain boundary, but a crystalline silicon film having no grain boundary is formed. It was almost impossible to form by laser annealing. Therefore, a TFT having a crystalline silicon film crystallized using a laser annealing method as an active layer and having characteristics equivalent to those of a MOS transistor manufactured on a single crystal silicon substrate has not been obtained to date.

  In order to form a high-quality crystalline semiconductor film with few defects, crystal grain boundaries, or crystal sub-boundaries on the insulating surface, a single crystal is known as a zone melting method. The mainstream method is to recrystallize after heating the semiconductor film on the substrate to a high temperature to bring it into a molten state.

  Since the underlying step is used as in the known graphoepitaxy technology, the crystal grows along the step, and the step remains on the surface of the formed single crystal semiconductor film. It has been. Furthermore, a single crystal semiconductor film could not be formed on a glass substrate with a relatively low strain point using graphoepitaxy.

  In any case, it has not been possible to form a crystalline semiconductor film free from defects, crystal grain boundaries, or subgrain boundaries due to volumetric shrinkage of the semiconductor caused by crystallization, thermal stress or lattice mismatch with the base. Therefore, without the bonding SOI (Silicon on Insulator), a crystalline semiconductor film formed on an insulating surface and crystallized or recrystallized can obtain the same quality as a MOS transistor formed on a single crystal substrate. I couldn't.

  The present invention has been made in view of the above problems, and forms a crystalline semiconductor film having at least a crystal grain boundary or a sub-grain boundary intersecting with the channel length direction as much as possible on an insulating surface at high speed. It is an object to provide a semiconductor device including a semiconductor element having high current driving capability.

  Further, it is possible to prevent the formation of a grain boundary in the channel formation region of the TFT, and to prevent the mobility of the TFT from being significantly lowered, the on-current is reduced, or the off-current is increased due to the grain boundary. An object is to provide a method for manufacturing a semiconductor device using a laser crystallization method and a semiconductor device manufactured using the manufacturing method.

  The inventors of the present invention formed a semiconductor film over an uneven insulating film and irradiated the laser film with a laser beam. When the semiconductor film was irradiated with laser light, the crystallized semiconductor film was selectively formed in a portion located on the convex portion of the insulating film. It was found that grain boundaries are formed.

  FIG. 19 shows the scanning direction of laser light when a non-single-crystal semiconductor film with a thickness of 200 nm formed over an uneven insulating film is irradiated with continuous-wave laser light at a scanning speed of 5 cm / sec. The cross-sectional image of TEM in a perpendicular direction is shown. In FIG. 19A, reference numerals 8001 and 8002 denote convex portions formed in the insulating film. The crystallized semiconductor film 8004 has a grain boundary 8003 above the convex portions 8001 and 8002.

  FIG. 19B schematically illustrates a cross-sectional image of the TEM illustrated in FIG. As shown in FIG. 19B, grain boundaries 8003 are formed on the upper portions of the convex portions 8001 and 8002. In the present inventors, this is because the semiconductor film is primarily melted by laser light irradiation, so that the semiconductor film located at the top of the insulating film moves in a volume direction toward the bottom of the concave portion. It was thought that the semiconductor film located above the film became thinner and could not withstand the stress, resulting in grain boundaries. In the crystallized semiconductor film, a grain boundary is selectively formed at the upper part of the convex part, while a grain boundary is formed in the part located in the concave parts (areas indicated by dotted lines) 8001 and 8002. Hateful. In addition, a recessed part points out the recessed area | region in which the convex part is not formed.

  Therefore, the present inventors thought that the position where the grain boundary is formed can be selectively determined by intentionally forming a portion where stress is concentrated on the semiconductor film. In the present invention, an insulating film having projections and depressions is formed on a substrate, and a semiconductor film is formed on the insulating film, so that stress is concentrated on the semiconductor film during crystallization by laser light. Are selectively formed. Specifically, unevenness is provided in the semiconductor film. Then, continuous oscillation laser light is irradiated along the longitudinal direction of the unevenness formed in the semiconductor film. At this time, it is most preferable to use a continuous wave laser beam, but a pulsed laser beam may be used. The cross section of the convex portion in the direction perpendicular to the scanning direction of the laser beam may be a rectangle, a triangle, or a trapezoid.

  With the above structure, grain boundaries are selectively formed on the convex portions of the semiconductor film during crystallization by laser light irradiation. The semiconductor film located on the recessed portion of the insulating film is relatively difficult to form grain boundaries. The semiconductor film located on the recessed portion of the insulating film has excellent crystallinity, but does not necessarily include the grain boundaries. However, even if a grain boundary exists, it can be said that the crystal grains are large and the crystallinity is relatively excellent as compared with the semiconductor film positioned on the convex portion of the insulating film. Therefore, at the stage where the shape of the insulating film is designed, the position where the grain boundary of the semiconductor film is formed can be predicted to some extent. That is, according to the present invention, the position where the grain boundary is formed can be selectively determined, so that it is possible to lay out the active layer so that the grain boundary is not included in the active layer, more preferably the channel formation region. Become.

  In the present invention, the semiconductor film located on the recess of the insulating film is positively used as the active layer of the TFT, so that it is possible to prevent the formation of a grain boundary in the TFT channel formation region. It is possible to prevent the mobility of the TFT from being significantly lowered, the on-current is reduced, and the off-current is increased. It should be noted that the designer can appropriately determine how far is removed by patterning in the vicinity of the edge of the convex portion or the concave portion.

  In order to solve the above problems, the present invention provides an insulating film in which an opening is provided over a substrate having an insulating surface, and a non-single-crystal semiconductor film is formed in the insulating film and the opening. The crystalline semiconductor film is formed by filling the opening by melting and crystallizing or recrystallizing the crystalline semiconductor film. The crystalline semiconductor film filling the opening, the gate electrode, Are characterized in that each step is formed so as to overlap with the gate insulating film interposed therebetween.

  The opening may be formed by etching the surface of the insulating substrate, or may be formed by etching the silicon oxide, silicon nitride, or silicon oxynitride film. The opening should be formed in accordance with the arrangement of the island-shaped semiconductor region including the channel formation region of the thin film transistor, and is desirably formed so as to match at least the channel formation region.

  As the non-single-crystal semiconductor film, an amorphous semiconductor film or a polycrystalline semiconductor film formed by a plasma CVD method, a sputtering method, or a low pressure CVD method, a polycrystalline semiconductor film formed by solid phase growth, or the like is applied. . Note that the amorphous semiconductor film referred to in the present invention is not limited to a film having a completely amorphous structure in a narrow sense, but includes a state in which fine crystal particles are included, or a so-called microcrystalline semiconductor film, local Includes a semiconductor film including a crystal structure. Typically, an amorphous silicon film is applied, and in addition, an amorphous silicon germanium film, an amorphous silicon carbide film, or the like can also be applied.

As a means for melting and crystallizing the non-single crystal semiconductor film, pulsed oscillation or continuous oscillation laser light using a gas laser oscillation device or a solid laser oscillation device as a light source is applied. The laser beam to be irradiated is condensed in a linear shape by an optical system, and its intensity distribution may have a uniform region in the longitudinal direction and may have a distribution in the lateral direction. As the oscillation device, a rectangular beam solid-state laser oscillation device is applied, and a slab laser oscillation device is particularly preferably applied. Alternatively, it is a solid-state laser oscillation device using a rod doped with Nd, Yb, Tm, and Ho, and in particular, a solid using a crystal doped with Nd, Tm, and Ho in a crystal such as YAG, YVO ₄ , YLF, and YAlO ₃ A laser oscillation device combined with a slab structure amplifier may be used. As the slab material, crystals such as Nd: YAG, Nd: GGG (gadolinium / gallium / garnet), Nd: GsGG (gadolinium / scandium / gallium / garnet) are used. A slab laser travels in a zigzag optical path through this plate-like laser medium while repeating total reflection.

  Moreover, you may irradiate the strong light according to it. For example, light having a high energy density obtained by collecting light emitted from a halogen lamp, xenon lamp, high-pressure mercury lamp, metal halide lamp, or excimer lamp with a reflecting mirror or lens may be used.

  The laser light or strong light focused and expanded linearly is applied to the non-single crystal semiconductor film, and the laser light irradiation position and the substrate on which the single crystal semiconductor film is formed are moved relative to each other. By scanning light partially or entirely, the non-single crystal semiconductor film is melted and crystallized or recrystallized. The scanning direction of the laser light is desirably performed along the longitudinal direction of the opening or the channel length direction. Thereby, a crystal grows along the scanning direction of the laser beam, and it is possible to prevent the crystal grain boundary or the crystal sub-grain boundary from intersecting the channel length direction. However, the present invention is not necessarily limited to this.

  In the semiconductor device of the present invention manufactured as described above, an insulating film having an opening formed on a substrate having an insulating surface is provided, and the crystalline semiconductor film formed on the substrate fills the opening. The channel region is provided in the filling region.

  In another structure, an insulating film in which an opening extending in the channel length direction is formed over a substrate having an insulating surface, and the crystalline semiconductor film formed on the substrate fills the opening. The channel region is provided in the filling region, and the opening has a depth equal to or greater than that of the crystalline semiconductor film.

  In another configuration, a crystalline semiconductor is provided in an opening formed in a rectangular shape or a strip shape formed on an insulating surface, and the crystalline semiconductor and the gate electrode overlap with each other through a gate insulating film. It is characterized by.

  In another configuration, a crystalline semiconductor is provided in an opening formed in the insulating surface and extending in the channel length direction, and the crystalline semiconductor and the gate electrode overlap with each other through the gate insulating film. It is characterized by.

  By setting the depth of the opening to be equal to or greater than the thickness of the semiconductor film, the semiconductor melted by the irradiation with laser light or strong light aggregates and solidifies into the opening (that is, the recess) due to surface tension. As a result, the thickness of the semiconductor film on the convex portion is reduced, and stress strain can be concentrated there. Further, the side surface of the opening has the effect of defining the crystal orientation to some extent. The angle of the side surface of the opening is 5 to 120 degrees, preferably 80 to 100 degrees with respect to the substrate surface.

  After the semiconductor film is melted by irradiation with laser light or strong light, solidification starts from a region where the bottom surface and side surface of the opening intersect, and crystal growth starts from here. For example, as shown in FIG. 17, in a system in which a step shape is formed by the insulating film (1) and the insulating film (2), a thermal analysis simulation at points A to D is performed. As a result, characteristics as shown in FIG. 18 are obtained. ing. Since there are both the insulating film (2) directly underneath and the insulating film (1) present on the side surface as a place for heat to escape, the temperature of the point B is the fastest. Thereafter, the order is A point, C point, and D point. This simulation result is for the case where the side wall angle is 45 degrees, but the same phenomenon can be considered qualitatively in the case of 90 degrees.

  In other words, the semiconductor film is once melted, aggregated into the opening formed on the insulating surface by surface tension, and crystal growth is caused from the approximate intersection of the bottom and side walls of the opening, thereby causing distortion caused by crystallization. It is possible to concentrate on other areas. That is, the crystalline semiconductor film formed so as to fill the opening can be freed from distortion.

  Note that the energy density in the vicinity of the edge of the laser beam of the laser light is generally lower than that in the vicinity of the center, and the crystallinity of the semiconductor film is often poor. For this reason, when scanning with laser light, it is desirable that the portion that will later become the channel formation region of the TFT and the edge of the locus do not overlap.

  Therefore, in the present invention, first, data (pattern information) of the shape of the insulating film or the semiconductor film viewed from the upper surface of the substrate, which is obtained at the design stage, is stored in the storage means. Then, from the pattern information and the width of the laser beam in the direction perpendicular to the scanning direction of the laser beam, at least the portion that becomes the channel formation region of the TFT and the edge of the laser beam trajectory do not overlap. Determine the scan path. Then, the position of the substrate is aligned with the marker as a reference, and the semiconductor film on the substrate is irradiated with laser light according to the determined scanning path.

  With the above-described configuration, it is possible to scan the laser beam at least at an indispensable portion instead of irradiating the entire substrate with the laser beam. Accordingly, it is possible to save time for irradiating the unnecessary portion with the laser beam, thereby shortening the time required for the laser beam irradiation and improving the processing speed of the substrate. Further, unnecessary portions can be irradiated with laser light to prevent the substrate from being damaged.

  Note that the marker may be formed by directly etching the substrate with a laser beam or the like, or the marker may be formed on a part of the insulating film at the same time when the insulating film having unevenness is formed. . In addition, the shape of the actually formed insulating film or semiconductor film is read using an imaging device such as a CCD, stored as data in the first storage means, and the insulation obtained at the design stage in the second storage means. The pattern information of the film or semiconductor film is stored, and the substrate is aligned by collating the data stored in the first storage unit with the pattern information stored in the second storage unit Anyway.

  By forming a marker on a part of the insulating film or using the shape of the insulating film as a marker, the number of marker masks can be reduced by one, and more accurately than a laser beam on the substrate. A marker can be formed, and the alignment accuracy can be improved.

  In general, the energy density of laser light is not completely uniform, and the height varies depending on the position in the laser beam. In the present invention, it is necessary to irradiate a laser beam having a certain energy density to a portion which becomes a channel forming region at least, more preferably, the entire flat surface of the concave portion or the entire flat surface of the convex portion. Therefore, in the present invention, by scanning with laser light, a region having a uniform energy density is completely overlapped with at least a portion that becomes a channel formation region, more preferably, the entire flat surface of the concave portion or the entire flat surface of the convex portion. It is necessary to use a laser beam having such an energy density distribution. In order to satisfy the above energy density condition, it is considered desirable to make the shape of the laser beam rectangular or linear.

  Further, a portion of the laser beam having a low energy density may be shielded through a slit. By using the slit, a laser beam having a relatively uniform energy density can be applied to the entire flat surface of the concave portion or the entire flat surface of the convex portion, and crystallization can be performed uniformly. Further, by providing the slit, the width of the laser beam can be partially changed depending on the pattern information of the insulating film or the semiconductor film, and restrictions on the layout of the channel formation region and further the active layer of the TFT can be reduced. Note that the width of the laser beam means the length of the laser beam in a direction perpendicular to the scanning direction.

  One laser beam obtained by combining laser beams oscillated from a plurality of laser oscillation devices may be used for laser crystallization. With the above configuration, it is possible to compensate for the low energy density portion of each laser beam.

  Further, after the semiconductor film is formed, laser light irradiation is performed so that the semiconductor film is not exposed to the air (for example, a specified gas atmosphere such as a rare gas, nitrogen, oxygen, or the like, or a reduced pressure atmosphere), and the semiconductor film is crystallized. Also good. With the above configuration, contaminants at the molecular level in the clean room, such as boron contained in a filter for increasing the cleanliness of air, can be prevented from being mixed into the semiconductor film during crystallization by laser light. it can.

The conventional semiconductor film crystallization technology called graphoepitaxy attempts to epitaxially grow a semiconductor film by induction of a surface relief grating on the surface of an artificially produced amorphous substrate. Met. The technology related to this graphoepitaxy is described in Non-Patent Document 2 and the like. In the above-mentioned papers and the like, the graphoepitaxy technology means that a step is provided on the surface of an insulating film, and the semiconductor film formed on the insulating film is subjected to a treatment such as heating or irradiation with laser light. Epitaxial growth of film crystals is disclosed. However, the temperature required for the epitaxial growth is required to be at least about 700 ° C. When the epitaxial growth is performed on the glass substrate, a grain boundary is formed in the semiconductor film in the vicinity of the edge of the concave portion or the convex portion of the insulating film. In the present invention, the shape of the concave portion or convex portion of the insulating film and the position of the edge are designed in accordance with the layout of the island so that the island mask is laid out and the crystallinity in the portion to be the island can be improved. Specifically, the shape, size, and the like of the concave portion or the convex portion are determined so that the edge of the concave portion or the convex portion and the center between the edges of the concave portion or the convex portion do not overlap with the island. Then, using the insulating film designed in accordance with the island layout, the position of the grain boundary is selectively determined intentionally. Then, a part where the selectively formed grain boundary exists in the semiconductor film is removed by patterning, and a part with relatively excellent crystallinity is used as an island. Therefore, the technique disclosed in the present invention is consistent with conventional graphoepitaxy in that a semiconductor film is formed on an insulating film provided with a step and the semiconductor film is crystallized using the step. This graphoepitaxy does not include the concept of controlling the position of grain boundaries using steps and reducing the grain boundaries in the island, and is similar to the present invention.

  By melt crystallization so that the opening formed on the insulating surface is filled with a semiconductor, strain generated by crystallization can be concentrated in a region other than the opening. That is, the crystalline semiconductor film formed so as to fill the opening can be freed from distortion.

  That is, in crystallization in which continuous wave laser light is irradiated to a non-single-crystal semiconductor film, an opening is formed in the base insulating film, and the opening is filled with a semiconductor in the process of melt crystallization. It is possible to concentrate strains associated with crystallization, crystal grain boundaries, or crystal sub-grain boundaries in regions other than the openings. By forming the TFT so that the channel formation region is disposed in the crystalline semiconductor film in the opening, the current driving capability can be improved at high speed, and the reliability of the element can also be improved. It becomes.

  In the present invention, the semiconductor film located on the recess of the base film is positively used as the active layer of the TFT, so that it is possible to prevent the formation of a grain boundary in the TFT channel formation region. It is possible to prevent the mobility of the TFT from being significantly lowered, the on-current is reduced, and the off-current is increased. It should be noted that the designer can appropriately determine how far is removed by patterning in the vicinity of the edge of the convex portion or the concave portion.

  In addition, since the plurality of channel formation regions of the TFT are separated from each other, a region overlapping with the gate electrode with the gate insulating film interposed therebetween can be widened in the channel formation region. Can be long. The heat generated by driving the TFT can be efficiently radiated while securing the on-current by increasing the channel width.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. The perspective view shown in FIG. 1 shows a form in which the first insulating film 102 and the second insulating films 103 to 105 patterned in a strip shape are formed on the substrate 101. Here, three strip-like patterns are formed by the second insulating film, but the number is not limited to that. As the substrate, a commercially available non-alkali glass substrate, a quartz substrate, a sapphire substrate, a substrate obtained by coating the surface of a single crystal or polycrystalline semiconductor substrate with an insulating film, or a substrate obtained by coating the surface of a metal substrate with an insulating film can be applied.

  The width W1 of the second insulating film formed in a strip shape is 0.1 to 10 μm (preferably 0.5 to 1 μm), and the interval W2 between adjacent second insulating films is 0.1 to 5 μm (preferably 0.5 to 1 μm). The thickness d of the second insulating film is the same as or greater than the thickness of the non-single-crystal semiconductor film formed thereon. Further, the step shape does not need to be a regular periodic pattern, and may be formed in accordance with the arrangement and shape of the island-shaped semiconductor regions including the channel formation region of the TFT. Accordingly, the length L of the second insulating film is not limited, and may be any length that can form a channel formation region of the TFT, for example.

The first insulating film is formed using silicon nitride or silicon nitride oxide. The second insulating film is formed using silicon oxide or silicon oxynitride. Silicon oxide can be formed by plasma CVD using a mixture of tetraethyl orthosilicate (TEOS) and O ₂ . The silicon nitride oxide film can be formed by a plasma CVD method using SiH ₄ , NH ₃ , N ₂ O, or SiH ₄ , N ₂ O as a raw material.

  As shown in FIG. 1, when the concave and convex shape by the opening is formed by the first insulating film and the second insulating film, the etching rate of the second insulating film is set to ensure the selection ratio in the etching process. It is desirable to appropriately adjust the material and the film formation conditions so as to be relatively fast. Then, the angle of the side wall of the opening formed by the second insulating film may be appropriately set in the range of 5 to 120 degrees, preferably 80 to 100 degrees.

  As shown in FIG. 2, an amorphous semiconductor film 106 is formed to a thickness of 50 to 200 nm covering the surface composed of the first insulating film 102 and the second insulating films 103 to 105 and the opening. For the amorphous semiconductor film, silicon, a compound or alloy of silicon and germanium, or a compound or alloy of silicon and carbon can be used.

Then, the amorphous semiconductor film 106 is crystallized by irradiating it with continuous wave laser light. The applied laser light is linearly condensed and expanded by an optical system, and its intensity distribution may have a uniform region in the longitudinal direction and may have a distribution in the short direction. As the laser oscillation device used as a rectangular beam solid-state laser oscillation device, a slab laser oscillation device is particularly preferably applied. Alternatively, it is a solid-state laser oscillation device using a rod doped with Nd, Tm, and Ho, and in particular a solid-state laser oscillation using a crystal doped with Nd, Tm, and Ho in a crystal such as YAG, YVO ₄ , YLF, and YAlO ₃ A device in which a slab structure amplifier is combined may be used. Then, as indicated by the arrows in the figure, scanning is performed in a direction intersecting the linear longitudinal direction. At this time, it is most desirable to scan in a direction parallel to the longitudinal direction of the belt-like pattern formed on the base insulating film. The term “linear” as used herein means that the ratio of the length in the longitudinal direction to the length in the short direction is 1 to 10 or more.

  As the slab material, crystals such as Nd: YAG, Nd: GGG (gadolinium / gallium / garnet), Nd: GsGG (gadolinium / scandium / gallium / garnet) are used. A slab laser travels in a zigzag optical path through this plate-like laser medium while repeating total reflection.

The wavelength of the continuous wave laser light is preferably 400 to 700 nm in consideration of the light absorption coefficient of the amorphous semiconductor film. Light in such a wavelength band is obtained by taking out the second harmonic and the third harmonic of the fundamental wave using a wavelength conversion element. As wavelength conversion elements, ADP (ammonium dihydrogen phosphate), Ba ₂ NaNb ₅ O ₁₅ (barium sodium niobate), CdSe (selenium cadmium), KDP (potassium dihydrogen phosphate), LiNbO ₃ (lithium niobate) , Se, Te, LBO, BBO, KB5, etc. are applied. It is particularly desirable to use LBO. As a typical example, the second harmonic (532 nm) of an Nd: YVO ₄ laser oscillation device (fundamental wave 1064 nm) is used. The laser oscillation mode is a single mode which is a TEM ₀₀ mode.

In the case of silicon selected as the most suitable material, the region having an absorption coefficient of 10 ³ to 10 ⁴ cm ⁻¹ is almost in the visible light region. When crystallizing a substrate having a high visible light transmittance such as glass and an amorphous semiconductor film formed with a thickness of 30 to 200 nm using silicon, by irradiating light in the visible light region with a wavelength of 400 to 700 nm, Crystallization can be performed without selectively damaging the base insulating film by selectively heating the semiconductor region. Specifically, the penetration depth of light having a wavelength of 532 nm is approximately 100 nm to 1000 nm with respect to the amorphous silicon film, and can sufficiently reach the inside of the amorphous semiconductor film 106 formed with a film thickness of 30 nm to 200 nm. it can. That is, heating can be performed from the inside of the semiconductor film, and almost the entire semiconductor film in the laser light irradiation region can be heated uniformly.

  The semiconductor melted by the laser light irradiation collects in the opening (concave portion) due to the surface tension. In the solidified state, the surface becomes almost flat as shown in FIG. Further, a crystal growth edge, a crystal grain boundary, or a crystal sub-grain boundary is formed on the second insulating film (on the convex portion) (region 110 indicated by hatching in the drawing). Thus, the crystalline semiconductor film 107 is formed.

  Thereafter, as shown in FIG. 4, the crystalline semiconductor film 107 is etched to form island-shaped semiconductor regions 108 and 109. At this time, only a high-quality semiconductor region can be left by etching away the region 110 where the growth edge, crystal grain boundary, or crystal sub-grain boundary is concentrated. Then, a gate insulating film and a gate electrode are formed using the crystalline semiconductor filling the opening (concave portion) of the island-shaped semiconductor regions 108 and 109 so that the channel formation region is positioned. A TFT can be completed through these steps.

  FIG. 5 is a diagram schematically illustrating the relationship between the depth and interval of the groove (step) of the opening formed by the first insulating film 102 and the second insulating films 103 to 105 and crystal growth. 5, t01: the thickness of the amorphous semiconductor film on the second insulating film (convex portion), t02: the thickness of the amorphous semiconductor film on the opening (concave portion), t11: thickness of the crystalline semiconductor film on the second insulating film (convex portion), t12: thickness of the crystalline semiconductor film on the opening portion (concave portion), d: thickness of the second insulating film (depth of the opening portion) W1: the width of the second insulating film, W2: the width of the opening.

  FIG. 5A shows the case of d <t02, W1, W2 ≦ 1 μm. When the depth of the groove in the opening is smaller than that of the amorphous semiconductor film 106, the melt crystallization process is performed. The opening is not filled with a semiconductor, and the surface of the crystalline semiconductor film is not flattened. That is, the uneven shape of the base of the crystalline semiconductor film remains roughly preserved.

  FIG. 5B shows the case of d ≧ t02, W1, W2 ≦ 1 μm. When the groove depth of the opening is substantially equal to or larger than that of the amorphous semiconductor film 106, the surface tension works. Gather in the opening (recess). In the solidified state, the surface becomes almost flat as shown in FIG. In this case, t11 <t12, and stress concentrates on the thin portion 120, strain is accumulated therein, and a crystal growth end is formed.

  FIG. 5C shows the case of d >> t02, W1, W2 ≦ 1 μm. In this case, the crystalline semiconductor film 107 is formed so as to fill the opening, and almost remains on the second insulating film. do not do.

  FIG. 5D shows the case of d ≧ t02 and W1, W2> 1 μm. When the width of the opening is widened, the crystalline semiconductor film fills the opening and has a flattening effect. Near the center, a grain boundary and a sub-grain boundary are generated. Similarly, stress is concentrated on the second insulating film, strain is accumulated therein, and a crystal growth edge is formed. This is presumed to be because the effect of stress relaxation is reduced by increasing the interval.

  FIG. 5E shows the case of d ≧ t02, W1, W2 >> 1 μm, and the state of FIG.

  As described above with reference to FIG. 5, it is considered that the conditions shown in FIG. 5B are most suitable when a semiconductor element is formed, particularly when a TFT is formed.

  In the above description, the example of the uneven shape of the base on which the crystalline semiconductor film is formed is formed using the first insulating film and the second insulating film, but is not limited to the form shown here and has the same shape If it is good. For example, the concave and convex shape may be provided by forming an opening directly on the surface of the quartz substrate.

  FIG. 6 shows an example of the configuration of a laser processing apparatus that can be applied for crystallization. FIG. 6 shows a laser oscillation device 301, a shutter 302, high conversion mirrors 303 to 306, a slit 307, cylindrical lenses 308 and 309, a mounting table 311, driving means 312 and 313 for displacing the mounting table 311 in the X and Y directions, A front view and a side view of the configuration of a laser processing apparatus comprising a control means 314 for controlling the drive means and an information processing means 315 for sending a signal to the laser oscillation apparatus 301 and the control means 314 based on a prestored program. It is shown.

  Laser light condensed linearly in the cross-sectional shape of the irradiation surface by the cylindrical lenses 308 and 309 is incident on the surface of the substrate 320 on the mounting table 311 obliquely. This is because the focal position shifts due to aberrations such as astigmatism, and a linear condensing surface can be formed at or near the irradiated surface. If the cylindrical lenses 308 and 309 are made of synthetic quartz, a high transmittance can be obtained, and the coating applied to the surface of the lens is applied in order to realize a transmittance of 99% or more with respect to the wavelength of the laser beam. Of course, the cross-sectional shape of the irradiation surface is not limited to a linear shape, and may be an arbitrary shape such as a rectangular shape, an elliptical shape, or an oval shape. In any case, the ratio of the short axis to the long axis is included in the range of 1:10 to 1: 100. Further, the wavelength conversion element 310 is provided for obtaining a harmonic with respect to the fundamental wave.

As described above, a rectangular beam solid-state laser oscillation device is applied as the laser oscillation device, and a slab laser oscillation device is particularly preferably applied. Alternatively, a slab structure amplifier may be combined with a solid-state laser oscillation device using a crystal in which Nd, Tm, and Ho are doped in a crystal such as YAG, YVO ₄ , YLF, and YAlO ₃ . As the slab material, crystals such as Nd: YAG, Nd: GGG (gadolinium / gallium / garnet), Nd: GsGG (gadolinium / scandium / gallium / garnet) are used. In addition, a gas laser oscillation device and a solid-state laser oscillation device capable of continuous oscillation can also be applied. As the continuous wave solid-state laser oscillator, a laser oscillator using a crystal in which Cr, Nd, Er, Ho, Ce, Co, Ti, or Tm is doped into a crystal such as YAG, YVO ₄ , YLF, or YAlO ₃ is applied.
The fundamental wave of the oscillation wavelength varies depending on the material to be doped, but oscillates at a wavelength of 1 μm to 2 μm. In order to obtain a higher output, a diode-pumped solid state laser oscillation device may be applied and cascaded.

  Further, the substrate 320 can be laser-processed by moving the mounting table 311 in the biaxial direction by the driving units 312 and 313. The movement in one direction can be continuously performed at a constant speed of 1 to 200 cm / sec, preferably 5 to 50 cm / sec, over a distance longer than the length of one side of the substrate 320, and the other side is a line. It is possible to discontinuously step the distance as much as the longitudinal direction of the beam. The oscillation of the laser oscillation device 301 and the mounting table 311 are operated in synchronism with an information processing means 315 equipped with a microprocessor.

  The mounting table 311 linearly moves in the X direction shown in the drawing, so that the entire surface of the substrate can be processed with a laser beam emitted from a fixed optical system. The position detection means 316 detects that the substrate 320 is at the irradiation position of the laser beam, transmits the signal to the information processing means 315, and synchronizes the timing with the oscillation operation of the laser oscillation device 301 by the information processing means 315. ing. In other words, when the substrate 320 is not at the irradiation position of the laser beam, the laser oscillation is stopped and the lifetime is extended.

  The laser light applied to the substrate 320 by the laser irradiation apparatus having such a structure can process a desired region or the entire surface of the semiconductor film by relatively moving in the X direction or the Y direction shown in the drawing.

  As described above, in crystallization of irradiating the amorphous semiconductor film with continuous wave laser light, by providing a stepped shape in the base insulating film, strain or stress accompanying crystallization can be concentrated on that portion, The crystalline semiconductor used as the active layer can be prevented from being strained or stressed. By forming a TFT so that a channel formation region is disposed in a crystalline semiconductor film free from strain or stress, it becomes possible to improve current drive capability at high speed and improve device reliability. Is also possible.

  In addition, a laser light irradiation method used in the present invention will be described with reference to FIG.

First, as shown in FIG. 20A, a first base film 9101 made of an insulating film is formed over a substrate. Then, a second base film 9102 made of a rectangular insulating film is formed over the first base film 9101, and the third base film 9103 is formed so as to cover the first and second base films 9101 and 9102. Is formed. In this embodiment mode, silicon nitride is used as the first base film 9101, silicon oxide is used as the second base film 9102, and a silicon oxide film is used as the third insulating film 9103. Note that the material of the first to third base films 9101 to 9103 is not limited to this, and is an alkali that can withstand heat treatment in a later step and can adversely affect the characteristics of the TFT in a semiconductor film to be formed later. Any insulating film can be used as long as it can prevent metal from being mixed and can form unevenness. The method for forming the unevenness will be described in detail later. Further, these other insulating films may be used. Alternatively, a stacked structure of two or more films may be used.

  In FIG. 20, the first to third base films are distinguished from each other, but the three base films are collectively referred to as a base film 9104. Note that although the base film 9104 having unevenness is formed using three base films in this embodiment mode, the structure of the base film used in the present invention is not limited thereto.

  At this time, the marker may be formed using a part of the base film simultaneously with the base film 9104.

  The substrate (not shown) may be any material that can withstand the processing temperature of the subsequent process. For example, a quartz substrate, a silicon substrate, a glass substrate such as barium borosilicate glass or aluminoborosilicate glass, a metal substrate, or a stainless substrate. A substrate having an insulating film formed on the surface can be used. Further, a plastic substrate having heat resistance enough to withstand the processing temperature may be used.

  Next, a semiconductor film 9105 is formed so as to cover the base film 9104. The semiconductor film 9105 can be formed by a known means (a sputtering method, an LPCVD method, a plasma CVD method, or the like). Note that the semiconductor film may be an amorphous semiconductor film, a microcrystalline semiconductor film, or a crystalline semiconductor film. In addition to silicon, silicon germanium may be used. In addition, after the third base film 9103 is formed, the film is continuously formed without being exposed to the atmosphere, so that impurities can be prevented from being mixed between the semiconductor film and the base film.

  Note that the shape and size of the convex portion of the base film 9104 can be set as appropriate by the designer, but the semiconductor film to be formed later is set to a thickness that does not cause film breakage in the vicinity of the edge of the convex portion. There is a need to.

  Next, as illustrated in FIG. 20B, the semiconductor film 9105 is irradiated with laser light. By irradiation with the laser light, the semiconductor film 9105 is temporarily melted, and its volume moves from the upper part of the convex part toward the concave part as indicated by the white arrow. Then, a semiconductor film (after LC) 9106 having a planarized surface and improved crystallinity is formed. The energy density of the laser beam is low in the vicinity of the edge of the laser beam. Therefore, crystal grains are small in the vicinity of the edge, and a protruding portion (ridge) appears along the crystal grain boundary. Therefore, the irradiation is performed so that the edge of the locus of the laser beam of the laser light does not overlap with the portion that becomes the channel formation region or the portion that is located on the concave portion of the semiconductor film 9105.

  Note that the scanning direction of the laser light is determined to be parallel to the longitudinal direction of the convex portion of the base film 9104.

In the present invention, a known laser can be used. Although it is desirable that the laser light be continuous oscillation, it is considered that the effect of the present invention can be obtained to some extent even if it is pulse oscillation. As the laser, a gas laser or a solid-state laser can be used. Examples of the gas laser include an excimer laser, an Ar laser, and a Kr laser. As the solid laser, a YAG laser, a YVO ₄ laser, a YLF laser, a YAlO ₃ laser, a glass laser, a ruby laser, an alexandrite laser, a Ti: sapphire laser, and a Y ₂ laser. For example, an O ₃ laser may be used. As the solid-state laser, a laser using a crystal such as YAG, YVO ₄ , YLF, or YAlO ₃ doped with Cr, Nd, Er, Ho, Ce, Co, Ti, Yb, or Tm is applied. The fundamental wave of the laser differs depending on the material to be doped, and a laser beam having a fundamental wave of about 1 μm can be obtained. The harmonic with respect to the fundamental wave can be obtained by using a nonlinear optical element.

  Furthermore, after converting infrared laser light emitted from a solid-state laser into green laser light by a nonlinear optical element, ultraviolet laser light obtained by another nonlinear optical element can also be used.

  The semiconductor film (after LC) 9106 has a large thickness on the concave portion of the base film 9104 and conversely has a thin film thickness on the convex portion due to volume movement due to laser light irradiation. Therefore, a grain boundary 9149 is likely to be generated on the convex portion due to the stress, and on the contrary, a substantially crystalline state is obtained on the concave portion. Note that the semiconductor film (after LC) 9106 does not necessarily include a grain boundary over the recess. However, even if there are grain boundaries, the crystal grains are large, so that the crystallinity is relatively excellent.

  Next, the surface of the semiconductor film (after LC) 9106 is etched to expose the upper surface of the convex portion of the base film 9104. Note that in this embodiment mode, the semiconductor film (after LC) 9106 is etched so that the upper surface of the convex portion of the base film 9104 is exposed. Then, as shown in FIG. 20C, a crystalline semiconductor film (island) 9107 is formed in the recess of the base film 9104.

  By using the island obtained by the above-described series of steps as an active layer of the TFT, more preferably as a channel formation region of the TFT, it is possible to prevent the formation of a grain boundary in the TFT channel formation region. Therefore, it is possible to prevent the mobility of the TFT from being significantly reduced, the on-current is reduced, and the off-current is increased. It should be noted that the designer can appropriately determine the extent to be removed by patterning as the vicinity of the edge of the concave portion or convex portion.

  Examples of the present invention will be described below.

  This embodiment shows an example in which a crystalline silicon film is formed over a base insulating film having an opening, and a TFT in which a channel formation region is provided in a filling region filled in the opening is shown.

In FIG. 7, a first insulating film 602 made of a 100 nm silicon oxynitride film is formed on a glass substrate 601. A silicon oxide film is formed thereon, and a second insulating film 603 having a rectangular pattern is formed by photolithography. The silicon oxide film is deposited with a thickness of 150 nm by mixing TEOS and O ₂ by plasma CVD, discharging at a reaction pressure of 40 Pa, a substrate temperature of 400 ° C., and a high frequency (13.56 MHz) power density of 0.6 W / cm ^2. Then, openings 604a and 604b are formed by etching.

  7A is a top view, FIG. 7B is a longitudinal sectional view corresponding to the line AA ′, and FIG. 7C is a longitudinal sectional view corresponding to the line BB ′. Hereinafter, FIGS. 8 to 12 are treated similarly.

Then, as shown in FIG. 8, an amorphous silicon film 605 covering the first insulating film 602 and the second insulating film 603 is formed with a thickness of 150 nm. The amorphous silicon film 605 is formed by plasma CVD using SiH ₄ as a source gas.

Then, as shown in FIG. 9, it is crystallized by irradiation with continuous wave laser light. The crystallization condition is a continuous wave mode YVO ₄ laser oscillator, and its second harmonic (wavelength 532 nm) output 5.5 W is longitudinally 400 μm in the longitudinal direction and 50-100 μm in the lateral direction by the optical system. The light is condensed so as to have a uniform energy density distribution in the direction, and is crystallized by scanning at a speed of 50 cm / sec. A uniform energy density distribution does not exclude anything that is completely constant, but the allowable range in the energy density distribution is ± 20%. A laser processing apparatus having the configuration shown in FIG. 6 can be applied to such laser light irradiation. The laser beam condensed by the optical system may have a region where the intensity distribution is uniform in the longitudinal direction and the distribution in the lateral direction. Crystallization allows the intensity distribution to be formed in a uniform region in the longitudinal direction, thereby increasing the effectiveness of crystal growth in a direction parallel to the laser beam scanning direction.

  By irradiating laser light under these conditions, the amorphous silicon film is instantaneously melted and crystallization proceeds while the molten zone moves. The molten silicon is agglomerated and solidified in the opening (concave portion) due to surface tension. As a result, the crystalline semiconductor film 606 is formed so as to fill the openings 604a and 604b.

  Thereafter, as shown in FIG. 10, a mask pattern is formed so that the crystalline semiconductor film remains in at least the openings 604a and 604b, and etching is performed to form island-shaped semiconductor regions 607 and 608 including a channel formation region. To do.

  FIG. 11 shows a state where the gate insulating film 609 and the gate electrodes 610 and 611 are formed on the upper layer side of the semiconductor regions 607 and 608. As the gate insulating film 609, an 80 nm silicon oxide film may be formed by a plasma CVD method. The gate electrodes 610 and 611 are formed using tungsten or an alloy containing tungsten. With such a structure, a channel formation region can be provided in the island-shaped semiconductor region that fills the openings 604a and 604b.

  Thereafter, the TFT can be completed by appropriately forming the source and drain regions, the low-concentration drain region, and the like.

  Although formed in the same process as in the first embodiment, as shown in FIG. 12, the shape of the opening formed in the second insulating film 603 is formed by an elongated strip-shaped region and a region connected thereto. Then, an island-shaped semiconductor region 620 made of a crystalline silicon film is formed in accordance with the opening 604c, and a gate insulating film 621 and a gate electrode 622 are formed, thereby forming a single-gate / multi-channel TFT. be able to.

  In Example 2, the second insulating film is formed thicker than the amorphous semiconductor film, for example, at 350 nm, thereby forming an island-shaped semiconductor formed of a crystalline semiconductor film as shown in FIG. Region 620 can be completely embedded in opening 604d. When the gate insulating film 621 and the gate electrode 622 are formed in the same manner, a single-gate / multi-channel TFT can be formed.

  FIG. 14 shows another example of a single-gate / multi-channel TFT. The first insulating film 602, the second insulating film 603, the island-shaped semiconductor region 630, the gate insulating film 631, and the gate electrode 632 are formed over the substrate 601 in the same manner as in the first to third embodiments. 14 is different from the opening 604e formed by the second insulating film 603 in the periphery of the semiconductor region where the channel formation region is formed after the island-shaped semiconductor region 630 is formed. The second opening 625 is formed by removing the two insulating films.

  FIG. 14D shows an enlarged view of the vicinity of the channel formation region. A gate insulating film 631 is formed in contact with a side surface and an upper surface of the island-shaped semiconductor region 630, and a gate electrode 632 is formed so as to cover the gate insulating film 631. In this case, the channel formation region is formed in both the upper portion 634 and the side surface portion 635 of the semiconductor region 630. As a result, the depletion region can be increased, and the current drive capability of the TFT can be improved.

  The present invention can be applied to various semiconductor devices, and the mode of a display panel manufactured based on Examples 1 to 4 will be described.

  In FIG. 15, a substrate 900 includes a pixel portion 902, gate signal side driver circuits 901a and 901b, a data signal side driver circuit 901c, an input terminal 935, and a wiring or a wiring group 917. The seal pattern 940 may partially overlap with the gate signal side driving circuits 901a and 901b, the data signal side driving circuit 901c, and the wiring or wiring group 917 connecting the driving circuit portion and the input terminal 935. In this way, the area of the frame region (the peripheral region of the pixel portion) of the display panel can be reduced. An FPC 936 is fixed to the input terminal 935.

  Further, a chip 950 in which a microprocessor, a memory, a media processor / DSP (Digital Signal Processor), or the like is formed using the TFT of the present invention may be mounted. These functional circuits are formed with design rules different from those of the pixel portion 902, the gate signal side drive circuits 901a and 901b, and the data signal side drive circuit 901c. Specifically, a design rule of 1 μm or less is applied. The There is no limitation on the mounting method, and a COG method or the like is applied.

  The TFTs shown in Embodiments 1 to 4 can be applied as switching elements of the pixel portion 902 and as active elements constituting the gate signal side driver circuits 901a and 901b and the data signal side driver circuit 901c.

  FIG. 16 shows an example of the configuration of one pixel in the pixel portion 902, which includes TFTs 801 to 803. These are switching TFTs, reset TFTs, and driving TFTs for controlling light emitting elements and liquid crystal elements included in the pixels.

  The island-shaped semiconductor regions 812 to 814 including the channel forming regions of these TFTs are formed in accordance with the openings 809 to 811 of the base insulating film formed in the lower layer. The island-shaped semiconductor regions 812 to 814 can be formed based on Examples 1 to 5. Gate lines 815 to 817 are formed above the island-shaped semiconductor regions 812 to 814, and a signal line 818, a power supply line 819, other various wirings 820 and 821, and a pixel electrode 823 are provided through a passivation film and a planarization film. Is formed.

  Thus, the present invention can complete the display panel without any influence.

  In this embodiment, a manufacturing process of a so-called multi-channel TFT having a plurality of channel formation regions separated from each other used in the semiconductor device of the present invention will be described.

  First, as illustrated in FIG. 21A, a base film 9120 having a convex portion is formed over an insulating surface. Note that FIG. 21B is a cross-sectional view taken along A-A ′ in FIG. 21A, and FIG. 21C is a cross-sectional view taken along B-B ′ in FIG.

  In this example, a base film 9120 having the same structure as that shown in the embodiment mode is used. The base film 9120 includes three base films. First, a second base film 9122 made of rectangular silicon oxide is formed on a first base film 9121 made of silicon nitride. A third base film 9123 made of silicon oxide is formed so as to cover the base films 9121 and 9122. In this embodiment, the base film 9120 is formed of the first to third base films 9121 to 9123. The convex portion 9124 of the base film 9120 includes a rectangular second insulating film 9122 and a portion of the third insulating film 9123 that is in contact with the second insulating film 9122 instead of the first insulating film 9121. It consists of

  Note that the shape and size of the convex portion 9124 can be set as appropriate by the designer, but it is necessary to set the thickness so that a semiconductor film to be formed later does not cause film breakage in the vicinity of the edge of the convex portion 9124. There is. In this embodiment, the height of the convex portion 9124 is set to about 0.1 to 1 μm.

  Note that the distortion of the substrate directly affects the shape of the underlying film to be formed later. Since distortion of the base film causes disorder in the uniformity of crystallinity of a semiconductor film to be formed later, a chemical mechanical polishing method (CMP method) is performed on the surface of the substrate so that the difference in distortion is suppressed to 10 nm or less. It is preferable to heat-treat the substrate in advance before forming the base film so that the substrate is not distorted by heat treatment in a later step.

  Then, an amorphous semiconductor film 9125 is formed so as to cover the base film 9120. The non-single-crystal semiconductor film 9125 can be formed by a known means (a sputtering method, an LPCVD method, a plasma CVD method, or the like). In this embodiment, a 300 nm non-single-crystal semiconductor film 9125 is formed by a plasma CVD method.

Next, as illustrated in FIG. 22A, the non-single-crystal semiconductor film 9125 is irradiated with laser light to be crystallized. Note that FIG. 22B corresponds to a cross-sectional view taken along dashed line AA ′ in FIG. In this embodiment, continuous wave YVO ₄ laser was used and irradiation was performed at a scanning speed of 50 cm / sec. At this time, the scanning direction of the laser light is aligned with the same direction as the carrier moves in a channel formation region to be formed later. In this embodiment, as indicated by the white arrow, the laser beam is irradiated with the scanning direction aligned with the longitudinal direction of the rectangular projection 9124. By irradiation with laser light, the non-single-crystal semiconductor film 9125 is melted, and the volume of the non-single-crystal semiconductor film 9125 is transferred from the top of the convex portion 9124 to the concave portion, so that a crystalline semiconductor film 9126 is formed.

  Next, the sub-island 9127 is formed by patterning the crystalline semiconductor film 9126 as shown in FIG. Note that FIG. 23B corresponds to a cross-sectional view taken along dashed line A-A ′ in FIG. A part of the sub-island 9127 exists on the concave portion formed between the convex portions 9124. The channel formation region of the target multi-channel TFT is formed using a portion located on the concave portion of the crystalline semiconductor film 9126. Therefore, in consideration of the number of channel formation regions, the channel length, and the channel width, It is important to define the positional relationship between the sub island 9127 and the convex portion 9124.

  Next, as illustrated in FIG. 24A, the island 9128 is formed by removing the sub-island 9127 from the upper surface so as to expose the upper surface of the convex portion 9124 of the base film 9120. Note that FIG. 24B corresponds to a cross-sectional view taken along dashed line A-A ′ in FIG. The sub-island 9127 can be removed from the upper surface by any method, for example, by etching or by CMP.

  By removing the sub-island 9127 from the upper surface, a portion where the grain boundary on the convex portion 9124 exists is removed, and there is almost no grain boundary on the concave portion corresponding to the convex portion 9124. A semiconductor film having excellent crystallinity is left in a portion to be a channel formation region. Then, as shown in FIGS. 24A and 24B, slit-like islands 9128 are formed by separating only the channel formation region. Note that the portion serving as the source region or the drain region is less affected by the crystallinity of the semiconductor film on the TFT characteristics than the channel formation region. Therefore, even if the portion that becomes the source region or the drain region has less crystallinity than the portion that becomes the channel formation region, it does not matter so much.

  Next, as illustrated in FIG. 25A, a TFT is manufactured using an island 9128. Note that there are various TFT structures and manufacturing methods. 25B corresponds to a cross-sectional view taken along a broken line AA ′ in FIG. 25A, and FIG. 25C corresponds to a cross-sectional view taken along a broken line BB ′ in FIG. 26A corresponds to a cross-sectional view taken along broken line CC ′ in FIG. 25A, and FIG. 26B corresponds to a cross-sectional view taken along broken line DD ′ in FIG.

  A channel formation region 9130 included in the island 9128 overlaps with the gate electrode 9132 with the gate insulating film 9131 interposed therebetween. Further, the channel formation region 9130 is sandwiched between two impurity regions 9133 that are also included in the island 9128. Note that the two impurity regions 9133 function as a source region or a drain region.

  A first interlayer insulating film 9134 is formed so as to cover the island 9128, the gate insulating film 9131, and the gate electrode 9132. A second interlayer insulating film 9135 is formed so as to cover the first interlayer insulating film 9134. Note that the first interlayer insulating film 9134 is an inorganic insulating film, and impurities such as carbon included in the second interlayer insulating film 9135 can be prevented from entering the island 9128. The second interlayer insulating film 9135 is an organic resin film, and has an effect of planarizing the surface so that wirings to be formed later are not disconnected.

  A wiring 9136 connected to the impurity region 9133 through a contact hole formed in the gate insulating film 9131, the first interlayer insulating film 9134, and the second interlayer insulating film 9135 is provided in the second interlayer insulating film 9135. Formed on top.

  Through the above manufacturing process, a TFT having a plurality of channel formation regions separated from each other is completed. With such a configuration, heat generated by driving the TFT can be efficiently radiated.

  In the present invention, the structure of the TFT is not limited to that shown in FIG. Further, the number of channel formation regions is not limited to four, and the number of channel formation regions may be one or a number other than four.

Further, the structure of the TFT is not limited to the above structure, and for example, it may have a structure as shown in FIG. The TFT illustrated in FIG. 27A includes a gate electrode including two layers of conductive films 9140 and 9141. A sidewall 9142 made of an insulating film is formed so as to be in contact with the upper surface of the conductive film 9140 and the side surface of the conductive film 9141. For example, TaN can be used as the conductive film 9140, W can be used as the conductive film 9141, and SiO ₂ can be used as the sidewall 9142. The TFT illustrated in FIG. 27B includes a gate electrode including two conductive films 9144 and 9145. The conductive film 9144 overlaps part of the impurity region.

  In the above step, after the laser light irradiation or the crystalline silicon film is etched to such an extent that the convex portion 9124 of the base film is exposed, the semiconductor film is heated at 500 to 600 ° C. for about 1 to 60 minutes. The stress generated inside can be relieved.

  In the present invention, the semiconductor film located on the recess of the insulating film is positively used as the active layer of the TFT, so that it is possible to prevent the formation of a grain boundary in the TFT channel formation region. It is possible to prevent the mobility of the TFT from being significantly lowered, the on-current is reduced, and the off-current is increased.

  In this example, a method for manufacturing an island, which is different in order of steps from Example 6, will be described. In addition, Example 6 is referred for the detailed description of each process.

  As shown in FIG. 28A, first, a base film having a rectangular projection 9301 is formed, and a non-single-crystal semiconductor film 9302 is formed over the base film. Next, the non-single-crystal semiconductor film 9302 is irradiated with laser light, so that a crystalline semiconductor film 9303 is formed (FIG. 28B).

  Next, part of the crystalline semiconductor film 9303 is removed from the surface thereof until the upper surface of the convex portion 9301 is exposed. Note that in this embodiment, the removal is performed by etching, and the crystalline semiconductor film after the removal is here a crystalline semiconductor film (after etching) 9304 (FIG. 28C).

  Next, the crystalline semiconductor film (after etching) 9304 is patterned to form islands 9305 (FIG. 28D).

  Note that in the above step, after laser light irradiation, the crystalline semiconductor film is etched to the extent that the protrusions of the base film are exposed, or the island is formed, and then heated at 500 to 600 ° C. for about 1 to 60 minutes. Thus, the stress generated in the semiconductor film can be relaxed.

  By the above process, before the island is formed, the crystalline semiconductor film is etched to such an extent that the convex portion of the base film is exposed, thereby preventing the end portion and the side surface of the island from being partially removed by etching. Can do.

  In this example, an island manufacturing method in which the process order is different from those in Examples 6 and 7 will be described. In addition, Example 6 is referred for the detailed description of each process.

  As shown in FIG. 29A, first, a base film having a rectangular protrusion 9311 is formed, and a non-single-crystal semiconductor film 9312 is formed over the base film.

  Next, the non-single-crystal semiconductor film 9312 is patterned to form a sub-island 9313 (FIG. 29B).

  Next, the sub-island 9313 is irradiated with laser light to be crystallized. In this embodiment, the sub-island after crystallization is defined as a sub-island (after crystallization) 9314 (FIG. 29C).

  Next, a part of the sub-island (after crystallization) 9314 is removed from the surface thereof until the upper surface of the convex portion 9311 is exposed. Note that in this embodiment, removal is performed using etching to form an island 9315 (FIG. 29D).

  Note that in the above steps, after laser light irradiation or island formation, the stress generated in the semiconductor film can be reduced by heating at 500 to 600 ° C. for about 1 to 60 minutes.

  In this embodiment, an example in which a multi-channel TFT and a single-channel TFT having only one channel formation region are formed using a plurality of convex portions will be described.

  FIG. 30A shows a base film having a plurality of rectangular projections 9330. A TFT using an island formed over the base film is shown in FIG. In FIG. 30B, a multi-channel TFT 9331 having four channel formation regions, a multi-channel TFT 9332 having two channel formation regions, and a single-channel TFT 9333 are included.

  Each TFT is formed on a concave portion in which a channel formation region is located between the convex portions 9330. More preferably, it is desirable that the channel formation region and the LDD region are formed on a recess located between the protrusions 9330.

  This example can be implemented in combination with Examples 6-8.

  In this embodiment, an example in which a laser light irradiation process and a semiconductor film crystallization process using a catalyst are combined in crystallization of a semiconductor film will be described. In the case of using a catalyst element, it is desirable to use the techniques disclosed in Japanese Patent Application Laid-Open Nos. 7-130652 and 8-78329.

First, as illustrated in FIG. 31A, a non-single-crystal semiconductor film 9352 is formed over a base film 9351 having a protrusion 9350. Next, the non-single-crystal semiconductor film 9352 is crystallized using a catalyst element (FIG. 31B). For example, in the case of using the technique disclosed in Japanese Patent Laid-Open No. 7-130652, a nickel-containing layer 9353 is formed by applying a nickel acetate salt solution containing 10 ppm of nickel in terms of weight to the non-single-crystal semiconductor film 9352, and 500 After the dehydrogenation step at 1 ° C. for 1 hour, heat treatment is performed at 500 to 650 ° C. for 4 to 12 hours, for example, 550 ° C. for 8 hours, so that a crystalline semiconductor film 9354 with improved crystallinity is formed. In addition to nickel (Ni), usable catalyst elements include germanium (Ge), iron (Fe), palladium (Pd), tin (Sn), lead (Pb), cobalt (Co), platinum (Pt). ), Copper (Cu), gold (Au), or the like may be used.

  Then, a crystalline semiconductor film (after LC) 9355 with further improved crystallinity is formed from the crystalline semiconductor film (after NiSPC) 9354 crystallized by NiSPC by laser light irradiation (FIG. 31C). ). The crystalline semiconductor film (after LC) 9355 is primarily melted when irradiated with laser light, and moves in volume from the upper part of the convex part 9350 toward the concave part, so that the surface is flattened. And the film thickness is thin on the convex part 9350, The grain boundary 9356 is easy to be made with stress.

  Next, a step of gettering the catalytic element in the crystalline semiconductor film (after LC) 9355 will be described. Note that in this embodiment, gettering is performed after laser light irradiation, but may be performed after the crystalline semiconductor film (after LC) 9355 is etched.

  A barrier layer 9357 containing silicon as a main component is formed in the crystalline semiconductor film (after LC) 9355 (FIG. 31D). Note that the barrier layer 9357 may be an extremely thin layer, and may be a natural oxide film, or may be an oxide film that is oxidized by generating ozone by irradiation of ultraviolet rays in an atmosphere containing oxygen. The barrier layer 9357 may be an oxide film oxidized with a solution containing ozone used for surface treatment called hydro-cleaning performed for removing carbon, that is, organic substances. This barrier layer 9357 is mainly used as an etching stopper. Further, after this barrier layer 9357 is formed, channel doping may be performed, and then activated by irradiation with strong light.

Next, a first semiconductor film 9358 for gettering is formed over the barrier layer 9357. The first semiconductor film 9358 for gettering may be a semiconductor film having an amorphous structure or a semiconductor film having a crystal structure. The thickness of the first semiconductor film 9358 for gettering is 5 to 50 nm, preferably 10 to 20 nm. The first semiconductor film 9358 for gettering contains oxygen (concentration in SIMS analysis is 5 × 10 ¹⁸ atoms / cm ³ or more, preferably 1 × 10 ¹⁹ atoms / cm ³ or more) to improve the gettering efficiency. It is desirable to improve.

  Next, a second semiconductor film (gettering site) 9359 containing a rare gas element is formed over the first semiconductor film 9358 for gettering. The second semiconductor film 9359 for gettering may be a semiconductor film having an amorphous structure using a plasma CVD method, a low pressure thermal CVD method, or a sputtering method, or a semiconductor film having a crystal structure. Also good. The second semiconductor film may be a semiconductor film containing a rare gas element in the film formation stage, or a rare gas element may be added after the formation of the semiconductor film not containing the rare gas element. In this embodiment, the second semiconductor film 9359 for gettering containing a rare gas element is formed in the film formation stage, and then the second semiconductor film 9359 for gettering is formed by selectively adding a rare gas element. An example is shown. Further, the first semiconductor film and the second semiconductor film for gettering may be continuously formed without being exposed to the atmosphere. Further, the sum of the thickness of the first semiconductor film and the thickness of the second semiconductor film may be 30 to 200 nm, for example, 50 nm.

In this embodiment, a gap between the crystalline semiconductor film (after LC) 9355 and the second semiconductor film 9359 is provided by the first semiconductor film 9358 for gettering. At the time of gettering, impurity elements such as metals existing in the crystalline semiconductor film (after LC) 9355 tend to gather near the boundary of the gettering site. It is desirable to improve the gettering efficiency by moving the boundary of the gettering site away from the crystalline semiconductor film (after LC) 9355 by the one semiconductor film 9358. In addition, the first semiconductor film 9358 for gettering also has an effect of blocking the impurity element contained in the gettering site from diffusing and reaching the interface of the first semiconductor film during gettering. doing. In addition, the first semiconductor film 9358 for gettering includes a rare gas element,
It also has an effect of protecting the crystalline semiconductor film (after LC) 9355 from being damaged.

Next, gettering is performed. As a step of performing gettering, heat treatment may be performed in a nitrogen atmosphere at 450 to 800 ° C. for 1 to 24 hours, for example, at 550 ° C. for 14 hours. Moreover, you may irradiate strong light instead of heat processing. Further, the substrate may be heated by injecting heated gas. In this case, heating may be performed at 600 ° C. to 800 ° C., more preferably at 650 ° C. to 750 ° C. for 1 to 60 minutes. Time can be shortened. By this gettering, the impurity element moves to the second semiconductor film 9359 as indicated by an arrow in FIG. 31D, and the impurity contained in the crystalline semiconductor film (after LC) 9355 covered with the barrier layer 9357 Element removal or impurity element concentration reduction is performed. By this gettering, a crystalline semiconductor in which almost no impurity element is present, that is, the impurity element concentration in the film is 1 × 10 ¹⁸ atoms / cm ³ or less, preferably 1 × 10 ¹⁷ atoms / cm ³ or less. A film (after gettering) 9360 is formed.

  Next, the first semiconductor film 9358 for gettering and the second semiconductor film 9359 are selectively removed using the barrier layer 9357 as an etching stopper.

  Then, after removing the barrier layer 9357 by changing the etching conditions, the crystalline semiconductor film (after gettering) 9360 is etched to the extent that the upper surface of the convex portion 9350 is exposed, as shown in FIG. The crystalline semiconductor film 9361 is formed in the recess.

  Note that after applying a solution containing a catalytic element to the semiconductor film before crystallization, crystal growth may be performed by laser light irradiation instead of SPC. For the gettering, a technique described in JP-A-10-135468 or JP-A-10-135469 may be used.

  In this embodiment, gettering is performed after laser light irradiation, but the present invention is not limited to this configuration. Gettering may be performed after the etching shown in FIG.

  This example can be implemented in combination with Examples 6 to 9.

  Next, the structure of the laser irradiation apparatus used in the present invention will be described with reference to FIG. Reference numeral 9151 denotes a laser oscillation device. Although four laser oscillation devices are used in FIG. 32, the number of laser oscillation devices included in the laser irradiation device is not limited to this number.

  Note that the laser oscillation device 9151 may use a chiller 9152 to keep the temperature constant. The chiller 9152 is not necessarily provided. However, by keeping the temperature of the laser oscillation device 9151 constant, it is possible to prevent the energy of the laser beam to be output from varying depending on the temperature.

  Reference numeral 9154 denotes an optical system which can focus the laser light by changing the optical path output from the laser oscillation device 9151 or processing the shape of the laser beam. Further, in the laser irradiation apparatus of FIG. 32, the laser beams of the laser beams output from the plurality of laser oscillation apparatuses 9151 can be combined by the optical system 9154 by overlapping each other.

  Note that an AO modulator 9153 that changes the traveling direction of the laser light in a very short time may be provided in the optical path between the substrate 9156 that is an object to be processed and the laser oscillation device 9151. Further, an attenuator (light quantity adjustment filter) may be provided instead of the AO modulator 9153 to adjust the energy density of the laser light.

  Further, a means (energy density measuring means) 9165 for measuring the energy density of the laser beam output from the laser oscillation device 9151 is provided in the optical path between the substrate 9156 that is the object to be processed and the laser oscillation device 9151, and the measurement is performed. The computer 9160 may monitor the energy density with time. In this case, the output from the laser oscillation device 9151 may be increased so as to compensate for the attenuation of the energy density of the laser beam.

  The combined laser beam is applied to the substrate 9156 as an object to be processed through the slit 9155. The slit 9155 is preferably formed of a material that can block the laser beam and that is not deformed or damaged by the laser beam. The slit 9155 has a variable slit width, and the width of the laser beam can be changed according to the slit width.

  Note that the shape of the laser beam on the substrate 9156 of the laser light oscillated from the laser oscillation device 9151 without the slit 9155 differs depending on the type of laser, and can also be shaped by an optical system.

  The substrate 9156 is placed on the stage 9157. In FIG. 32, position control means 9158 and 9159 correspond to means for controlling the position of the laser beam on the workpiece, and the position of the stage 9157 is controlled by the position control means 9158 and 9159.

  In FIG. 32, the position control means 9158 controls the position of the stage 9157 in the X direction, and the position control means 9159 controls the position of the stage 9157 in the Y direction.

  32 has a computer 9160 having both a storage means such as a memory and a central processing unit. The computer 9160 controls the oscillation of the laser oscillation device 9151, determines the laser beam scanning path, and controls the position control means 9158 and 9159 so that the laser beam of the laser beam is scanned according to the determined scanning path. The substrate can be moved to a predetermined position.

  In FIG. 32, the position of the laser beam is controlled by moving the substrate, but it may be moved by using an optical system such as a galvanometer mirror, or both.

  Further, in FIG. 32, the width of the slit 9155 may be controlled by the computer 9160 so that the width of the laser beam can be changed in accordance with the mask pattern information. Note that the slit is not necessarily provided.

  Further, the laser irradiation apparatus may include means for adjusting the temperature of the object to be processed. Further, since the laser light is light having high directivity and energy density, a damper may be provided to prevent the reflected light from being irradiated to an inappropriate place. The damper desirably has a property of absorbing reflected light, and cooling water may be circulated in the damper to prevent the temperature of the partition wall from rising due to absorption of the reflected light. Further, the stage 9157 may be provided with means for heating the substrate (substrate heating means).

When the marker is formed by a laser, a marker laser oscillation device may be provided. In this case, the oscillation of the marker laser oscillation device may be controlled by the computer 9160. Further, when a marker laser oscillation device is provided, an optical system for condensing the laser light output from the marker laser oscillation device is separately provided. The laser used for forming the marker is typically a YAG laser, a CO ₂ laser, or the like, but of course, other lasers can be used.

  In addition, one CCD camera 9163, or in some cases, several may be provided for alignment using markers. The CCD camera means a camera using a CCD (charge coupled device) as an image sensor.

  In addition, without providing the marker, the CCD camera 9163 may recognize the pattern of the insulating film or the semiconductor film and align the substrate. In this case, the pattern information of the insulating film or semiconductor film by the mask input to the computer 9160 and the actual insulating film or semiconductor film pattern information collected by the CCD camera 9163 are collated to grasp the position information of the substrate. can do. In this case, it is not necessary to provide a marker separately.

  In addition, the laser light incident on the substrate is reflected by the surface of the substrate and returns to the same optical path as the incident light, which is so-called return light, but the return light is a change in laser output, frequency, rod breakage, etc. Adverse effects. Therefore, an isolator may be installed in order to remove the return light and stabilize the oscillation of the laser.

  Note that FIG. 32 shows the configuration of the laser irradiation apparatus provided with a plurality of laser oscillation apparatuses, but one laser oscillation apparatus may be provided. FIG. 33 shows a configuration of a laser irradiation apparatus having one laser oscillation apparatus. In FIG. 33, reference numeral 9201 denotes a laser oscillation device, and 9202 denotes a chiller. 9215 is an energy density measuring means, 9203 is an AO modulator, 9204 is an optical system, 9205 is a slit, and 9213 is a CCD camera. The substrate 9206 is placed on the stage 9207, and the position of the stage 9207 is controlled by the X direction position control means 9208 and the Y direction position control means 9209. 32, the operation of each unit included in the laser irradiation apparatus is controlled by the computer 9210. The difference from FIG. 32 is that there is one laser oscillation apparatus. Unlike the case of FIG. 32, the optical system 9204 only needs to have a function of condensing one laser beam.

  Instead of scanning and irradiating the entire semiconductor film with laser light, the semiconductor film is crystallized and removed by patterning by scanning the laser light so that at least the indispensable part can be crystallized at least. The time for irradiating the portion to be irradiated with laser light can be saved, and the processing time per substrate can be greatly shortened.

  This example can be implemented in combination with Examples 6 to 10.

  In this embodiment, a method for forming a base film having unevenness will be described.

First, as shown in FIG. 34A, a first base film 9251 made of an insulating film is formed over a substrate 9250. In this embodiment, silicon oxynitride is used for the first base film 9251; however, the present invention is not limited to this, and any insulating film having a high selectivity in etching with respect to the second base film may be used. In this embodiment, the first base film 9251 is formed with a CVD apparatus so as to have a thickness of 50 to 200 nm using SiH ₄ and N ₂ O. Note that the first base film may be a single layer or a structure in which a plurality of insulating films are stacked.

  Next, as shown in FIG. 34B, a second base film 9252 made of an insulating film is formed so as to be in contact with the first base film 9251. The second base film 9252 needs to be patterned in a later step so that when the unevenness is formed, the unevenness appears on the surface of the semiconductor film to be formed thereafter. In this embodiment, as the second base film 9252, silicon oxide with a thickness of 30 nm to 300 nm is formed by a plasma CVD method.

Next, a mask 9253 is formed as shown in FIG. 34C, and the second base film 9252 is etched. In this example, a mixed solution (product name: LAL500, manufactured by Stella Chemifa) containing 7.13% ammonium hydrogen fluoride (NH ₄ HF ₂ ) and 15.4% ammonium fluoride (NH ₄ F) was used as an etchant. And wet etching at 20 ° C. By this etching, a rectangular convex portion 9254 is formed. In this specification, the first base film 9251 and the protrusion 9254 are combined and regarded as one base film.

  Note that in the case where aluminum nitride, aluminum nitride oxide, or silicon nitride is used as the first base film 9251 and a silicon oxide film is used as the second base film 9252, the second base film 9252 is patterned by an RF sputtering method. It is desirable. Since aluminum nitride, aluminum nitride oxide, or silicon nitride as the first base film 9251 has high thermal conductivity, generated heat can be quickly diffused, and deterioration of the TFT can be prevented.

  Next, a semiconductor film is formed so as to cover the first base film 9251 and the convex portion 9254. In this embodiment, since the thickness of the convex portion is 30 nm to 300 nm, the thickness of the semiconductor film is desirably 50 to 200 nm, and here, 60 nm. Note that if an impurity is mixed between the semiconductor film and the base film, the crystallinity of the semiconductor film may be adversely affected, which may increase variations in characteristics of TFTs to be manufactured and variations in threshold voltage. The semiconductor film is preferably formed continuously. Therefore, in this embodiment, after the formation of the base film composed of the first base film 9251 and the convex portion 9254, the silicon oxide film 9255 is thinly formed on the base film and then continuously exposed not to the atmosphere. Thus, a semiconductor film 9256 is formed. The thickness of the silicon oxide film can be set as appropriate by the designer, but in this embodiment, the thickness is about 5 nm to 30 nm.

  Next, a method of forming a base film different from that in FIG. 34 will be described. First, as shown in FIG. 35A, a first base film 9261 made of an insulating film is formed over a substrate 9260. The first base film is formed using a silicon oxide film, a silicon nitride film, a silicon oxynitride film, or the like.

In the case of using a silicon oxide film, tetraethyl orthosilicate (TEOS) and O ₂ are mixed by plasma CVD to have a reaction pressure of 40 Pa, a substrate temperature of 300 to 400 ° C., and high frequency (13.56 MHz) power. It can be formed by discharging at a density of 0.5 to 0.8 W / cm ² . In the case of using a silicon oxynitride film, SiH _4, N ₂ O, by forming silicon oxynitride film, or SiH _4, N silicon oxynitride film made from ₂ O, made from NH ₃ by the plasma CVD method It ’s fine. The production conditions in this case are a reaction pressure of 20 to 200 Pa, a substrate temperature of 300 to 400 ° C., and a high frequency (60 MHz) power density of 0.1 to 1.0 W / cm ² . Alternatively, a silicon oxynitride silicon film formed from SiH ₄ , N ₂ O, and H ₂ may be used. Similarly, the silicon nitride film can be formed from SiH ₄ and NH ₃ by plasma CVD.

The first base film is formed over the entire surface of the substrate to a thickness of 20 to 200 nm (preferably 30 to 60 nm), and then a mask 9262 is formed using a photolithography technique as shown in FIG. Then, unnecessary portions are removed by etching, and a rectangular convex portion 9263 is formed. For the first base film 9261, a dry etching method using a fluorine-based gas may be used, or a wet etching method using a fluorine-based aqueous solution may be used. When the latter method is selected, for example, a mixed solution containing 7.13% ammonium hydrogen fluoride (NH ₄ HF ₂ ) and 15.4% ammonium fluoride (NH ₄ F) (manufactured by Stella Chemifa Corporation, Etching under the trade name LAL500) is preferable.

  Next, a second base film 9264 made of an insulating film is formed so as to cover the protrusions 9263 and the substrate 9260. This layer is formed of a silicon oxide film, a silicon nitride film, a silicon oxynitride film, or the like with a thickness of 50 to 300 nm (preferably 100 to 200 nm), like the first base film 9261.

  Through the manufacturing process, a base film including the convex portion 9263 and the second base film 9264 is formed. Note that after the second base film 9264 is formed, impurities in the air are mixed between the semiconductor film and the base film by continuously forming the semiconductor film so as not to be exposed to the air. Can be prevented.

  This embodiment can be implemented by freely combining with Embodiments 6-11.

  In this embodiment, the shape of a laser beam synthesized by superimposing a plurality of laser beams will be described.

FIG. 36A shows an example of the shape of a laser beam on the object to be processed in the case where the laser light emitted from each of the plurality of laser oscillation devices does not pass through the slit. The laser beam shown in FIG. 36A has an elliptical shape. In the present invention, the shape of the laser beam emitted from the laser oscillation device is not limited to an ellipse. The shape of the laser beam varies depending on the type of laser and can also be shaped by an optical system. For example, the shape of a laser beam emitted from a Lambda XeCl excimer laser (wavelength 308 nm, pulse width 30 ns) L3308 is a rectangular shape of 10 mm × 30 mm (both half-value width in the beam profile). Further, the shape of the laser light emitted from the YAG laser is circular when the rod shape is cylindrical, and is rectangular when the rod shape is slab type. By further shaping such laser light with an optical system, laser light of a desired size can be produced.

FIG. 36B shows the energy density distribution of the laser light in the long axis Y direction of the laser beam shown in FIG. The laser beam shown in FIG. 36A corresponds to a region satisfying an energy density of 1 / e ² of the peak value of the energy density in FIG. The energy density distribution of the laser beam having a laser beam having an elliptical shape becomes higher toward the center O of the ellipse. As described above, in the laser beam illustrated in FIG. 36A, the energy density in the central axis direction follows a Gaussian distribution, and a region where it can be determined that the energy density is uniform is narrowed.

  Next, FIG. 36C illustrates the shape of a laser beam when the laser light including the laser beam illustrated in FIG. Note that FIG. 36C illustrates the case where one linear laser beam is formed by superimposing four laser beams, but the number of superimposed laser beams is not limited thereto.

  As shown in FIG. 36C, the laser beams of the respective laser beams are synthesized by matching the long axes of the respective ellipses and overlapping a part of the laser beams to form one laser beam 9360. Yes. Hereinafter, a straight line obtained by connecting the centers O of the ellipses will be referred to as a central axis of the laser beam 9360.

FIG. 36D shows the energy density distribution of the laser light in the central axis y direction of the combined laser beam shown in FIG. Note that the laser beam illustrated in FIG. 36C corresponds to a region satisfying an energy density of 1 / e ² of the peak value of the energy density in FIG. The energy density is added at the portion where the laser beams before synthesis are overlapped. For example, when the energy densities E1 and E2 of the overlapping beams are added as shown in the figure, the energy density peak value E3 is approximately equal, and the energy density is flattened between the centers O of the ellipses.

  Note that when E1 and E2 are added, it is ideally equal to E3, but in reality, it is not necessarily equal. The allowable range of deviation between the value obtained by adding E1 and E2 and the value of E3 can be appropriately set by the designer.

  When the laser beam is used alone, the energy density distribution follows a Gaussian distribution, so that the entire semiconductor film or island part in contact with the flat part of the insulating film can be irradiated with laser light with a uniform energy density. difficult. However, as can be seen from FIG. 36D, by superimposing a plurality of laser beams to complement each other with low energy density, it is possible to save energy compared to using a plurality of laser beams without overlapping. A region having a uniform density is enlarged, and the crystallinity of the semiconductor film can be increased efficiently.

FIG. 37 shows energy density distributions along BB ′ and CC ′ in FIG. 36C obtained by calculation. Note that FIG. 37 is based on a region satisfying an energy density of 1 / e ² of the peak value of the laser beam before synthesis. When the length of the laser beam before synthesis is 37 μm, the length of the long axis is 410 μm, and the distance between centers is 192 μm, the energy density at BB ′ and CC ′ is as follows: The distributions are as shown in FIGS. 37A and 37B, respectively. BB ′ is smaller than CC ′, but it can be regarded as almost the same size, and satisfies the energy density of 1 / e ² of the peak value of the laser beam before synthesis. The shape of the synthesized laser beam in can be expressed as a linear shape.

FIG. 38A is a diagram showing the energy distribution of the combined laser beam. A region indicated by 9361 is a region having a uniform energy density, and a region indicated by 9362 is a region having a low energy density. In FIG. 38, the length in the central axis direction of the laser beam is W _TBW, and the length in the central axis direction in a region 9361 having a uniform energy density is W _max . As W _TBW becomes larger than W _max , the ratio of the region 9362 that has a non-uniform energy density that cannot be used for crystallization of the semiconductor film to the region 9361 that has a uniform energy density that can be used for crystallization. growing. In the semiconductor film irradiated only with the region 9362 where the energy density is not uniform, microcrystals are generated and the crystallinity is not good. Accordingly, it is necessary to determine the layout of the scanning path and the unevenness of the insulating film so that only the region 9362 and the region to be the island of the semiconductor film do not overlap, and the restriction becomes larger as the ratio of the region 9362 to the region 9361 increases. . Therefore, using the slits to prevent the semiconductor film formed on the concave portion or the convex portion of the insulating film from being irradiated with only the region 9362 having a non-uniform energy density can prevent the layout of the scanning path and the concave and convex portions of the insulating film. This is effective in reducing the constraints that occur in the process.

  This example can be implemented in combination with Examples 6-12.

  In this embodiment, the optical system of the laser irradiation apparatus used in the present invention and the positional relationship between each optical system and the slit will be described.

  Laser light having an elliptical laser beam has a rectangular or linear laser beam in which the energy density distribution in the direction perpendicular to the scanning direction follows a Gaussian distribution, so that the ratio of the low energy density region to the entire region is a rectangular or linear laser beam. Higher than light. Therefore, in the present invention, it is desirable that the laser beam of the laser beam is rectangular or linear with a relatively uniform energy density distribution.

  FIG. 39 shows an optical system in the case where four laser beams are combined into one laser beam. The optical system shown in FIG. 39 has six cylindrical lenses 9417-9422. The four laser beams incident from the directions of the arrows are incident on the four cylindrical lenses 9419 to 9422, respectively. The two laser beams formed by the cylindrical lenses 9419 and 9421 are irradiated again onto the object 9423 after the shape of the laser beam is formed again by the cylindrical lens 9417. On the other hand, the two laser beams shaped by the cylindrical lenses 9420 and 9422 are shaped again by the cylindrical lens 9418 and irradiated to the object 9423.

  The laser beams of the respective laser beams on the workpiece 9423 are combined by overlapping each other to form one laser beam.

  The focal length and incident angle of each lens can be set as appropriate by the designer, but the focal lengths of the cylindrical lenses 9417 and 9418 closest to the object 9423 are smaller than the focal lengths of the cylindrical lenses 9419 to 9422. To do. For example, the focal lengths of the cylindrical lenses 9417 and 9418 closest to the workpiece 9423 are 20 mm, and the focal lengths of the cylindrical lenses 9419 to 9422 are 150 mm. The incident angle of laser light from the cylindrical lenses 9417 and 9418 to the object 9400 is 25 ° in this embodiment, and the incident angle of laser light from the cylindrical lenses 9419 to 9422 to the cylindrical lenses 9417 and 9418 is 10 °. Install each lens as you want. In order to prevent return light and perform uniform irradiation, it is desirable to keep the incident angle of the laser light on the substrate larger than 0 °, preferably 5 to 30 °.

  FIG. 39 shows an example of synthesizing four laser beams. In this case, there are four cylindrical lenses respectively corresponding to the four laser oscillation devices and two cylindrical lenses corresponding to the four cylindrical lenses. ing. The number of laser beams to be combined is not limited to this, and the number of laser beams to be combined may be 2 or more and 8 or less. When combining n (n = 2, 4, 6, 8) laser beams, n cylindrical lenses respectively corresponding to the n laser oscillation devices, and n / 2 cylindrical lenses corresponding to the n cylindrical lenses, have. When combining n (n = 3, 5, 7) laser beams, n cylindrical lenses respectively corresponding to the n laser oscillation devices, and (n + 1) / 2 cylindrical lenses corresponding to the n cylindrical lenses, have.

  When superposing five or more laser beams, it is desirable to irradiate the fifth and subsequent laser beams from the opposite side of the substrate in consideration of the location where the optical system is disposed and interference, etc. It must also be provided on the opposite side. Further, the substrate needs to have transparency.

  In order to prevent the return light from returning along the original optical path, it is desirable to keep the incident angle with respect to the substrate larger than 0 and smaller than 90 °.

Also, in order to realize uniform laser light irradiation, a plane or long side that is a plane perpendicular to the irradiation surface and includes a short side when the shape of each beam before synthesis is regarded as a rectangle is obtained. If any one of the surfaces to be included is defined as an incident surface, the incident angle θ of the laser light is set such that the length of the short side or the long side included in the incident surface is W, and is set on the irradiation surface, and It is desirable that θ ≧ arctan (W / 2d) is satisfied when the thickness of the substrate having translucency with respect to the laser beam is d. This argument needs to hold for each laser beam before synthesis. When the locus of the laser beam is not on the incident surface, the incident angle of the projection of the locus onto the incident surface is defined as θ. If the laser light is incident at this incident angle θ, the reflected light from the surface of the substrate and the reflected light from the back surface of the substrate do not interfere with each other, and uniform laser light irradiation can be performed. In the above discussion, the refractive index of the substrate was considered as 1. Actually, in many cases, the refractive index of the substrate is around 1.5, and if this value is taken into consideration, a calculated value larger than the angle calculated in the above discussion can be obtained. However, since the energy at both ends in the longitudinal direction of the beam spot is attenuated, the influence of interference in this portion is small, and the effect of interference attenuation can be sufficiently obtained with the above calculated value. The above inequality for θ does not apply to anything other than the substrate being translucent to the laser beam.

  Note that the optical system included in the laser irradiation apparatus used in the present invention is not limited to the structure shown in this embodiment.

  An excimer laser is a typical gas laser that can obtain a rectangular or linear laser beam without combining a plurality of laser beams, and a slab laser is a typical solid-state laser. In the present invention, these lasers may be used. It is also possible to form a linear or rectangular laser beam with a uniform energy density using an optical fiber.

  This example can be implemented in combination with Examples 6 to 13.

  In this embodiment, the relationship between the distance between the centers of laser beams and the energy density when the laser beams are superimposed will be described.

  In FIG. 40, the energy density distribution in the central axis direction of each laser beam is indicated by a solid line, and the energy density distribution of the synthesized laser beam is indicated by a broken line. The value of the energy density in the central axis direction of the laser beam generally follows a Gaussian distribution.

Let X be the distance between peaks when the distance in the central axis direction satisfying the energy density of 1 / e ² or more of the peak value in the beam spot before synthesis is 1. In the combined beam spot, Y is the percent increment of the peak value after the combination and the average value of the valley values. The relationship between X and Y obtained by simulation is shown in FIG. In FIG. 41, Y is expressed as a percentage.

In FIG. 41, the energy difference Y is expressed by the following approximate expression (Equation 1).
(Equation 1)
Y = 60-293X + 340X ² (X is the larger of the two solutions)

  According to Equation 1, for example, when the energy difference is desired to be about 5%, it can be understood that X≈0.584. Ideally, Y = 0. However, since the length of the beam spot is shortened, it is preferable to determine X in balance with the throughput.

Next, the allowable range of Y will be described. FIG. 42 shows a distribution of the output (W) of the YVO ₄ laser with respect to the beam width in the central axis direction when the laser beam has an elliptical shape. The region indicated by hatching is the range of output energy necessary for obtaining good crystallinity, and it is understood that it is sufficient that the output energy of the synthesized laser beam falls within the range of 3.5 to 6 W.

  When the maximum value and the minimum value of the output energy of the beam spot after synthesis enter the limit of the output energy range necessary for obtaining good crystallinity, the energy difference Y for obtaining good crystallinity is maximized. Therefore, in the case of FIG. 42, the energy difference Y is ± 26.3%, and it can be seen that good crystallinity can be obtained if the energy difference Y is within the above range.

  Note that the range of output energy required to obtain good crystallinity varies depending on how far the crystallinity is judged to be good, and the distribution of output energy also varies depending on the shape of the laser beam. The allowable range of Y is not necessarily limited to the above value. The designer needs to appropriately determine the range of output energy necessary for obtaining good crystallinity, and set the allowable range of the energy difference Y from the distribution of the output energy of the laser used.

  This example can be implemented in combination with Examples 6-14.

  The present invention can be applied to various semiconductor devices. A display panel manufactured based on Examples 6 to 10 will be described with reference to FIGS. 43 and 44. FIG.

  43, a substrate 9901 is provided with a pixel portion 9902, gate signal side driver circuits 9901a and 9901b, a data signal side driver circuit 9901c, an input / output terminal portion 9908, and a wiring or a wiring group 9904. The shield pattern 9905 may partially overlap with the gate signal side driver circuits 9901a and 9901b, the data signal side driver circuit 9901c, and a wiring or a wiring group 9904 that connects the driver circuit portion and the input terminal. In this way, the area of the frame region (the peripheral region of the pixel portion) of the display panel can be reduced. An FPC 9903 is fixed to the input / output terminal portion 9908.

  The present invention can be used for active elements constituting the pixel portion 9902, the gate signal side driver circuits 9901a and 9901b, and the data signal side driver circuit 9901c.

  FIG. 44 shows an example of the structure of one pixel of the pixel portion 9902 shown in FIG. In this embodiment, a pixel of a light-emitting device which is one of the semiconductor devices of the present invention will be described. Note that the light emitting device is a general term for a display panel in which a light emitting element formed on a substrate is sealed between the substrate and a cover material, and a display module in which a TFT or the like is mounted on the display panel. Note that the light-emitting element includes a layer (light-emitting layer) containing an organic compound from which luminescence (Electro Luminescence) generated by applying an electric field is obtained, an anode, and a cathode.

  Note that the light-emitting element used in this example is formed using a material in which a hole injection layer, an electron injection layer, a hole transport layer, an electron transport layer, or the like is an inorganic compound alone or an organic compound is mixed with an inorganic compound. It can also take the form which is made. These layers may be partially mixed with each other.

  Reference numeral 9801 denotes a TFT (switching TFT) as a switching element that controls input of a video signal input to the pixel. Reference numeral 9802 denotes a TFT (driving TFT) for supplying current to the pixel electrode based on information included in the video signal. TFT).

  The switching TFT 9801 has an active layer 9803 having a channel width of about 1 to 2 μm and having a plurality of channel formation regions, a gate insulating film (not shown), and a gate electrode 9805 which is a part of the gate line 9804. doing. Switching of the switching TFT 9801 is controlled by a selection signal input to the gate line 9804 from the gate signal side driving circuits 9901a and 9901b.

  One of a source region and a drain region of the active layer 9803 of the switching TFT 9801 is connected to a signal line 9806 to which a video signal is input by the data signal side driver circuit 9901c, and the other is connected to a wiring 9807 for connecting elements. Yes.

  Reference numeral 9820 denotes a convex portion of the base film used when the active layer 9803 is formed.

  On the other hand, the driving TFT 9802 has an active layer 9808 having a channel width of about 1 to 2 μm and a plurality of channel formation regions, a gate insulating film (not shown), and a gate electrode 9810 which is a part of the capacitor wiring 9809. And have.

  One of a source region and a drain region of the active layer 9808 of the driving TFT 9802 is connected to the power supply line 9811 and the other is connected to the pixel electrode 9812.

  Reference numeral 9821 denotes a convex portion of the base film used when the active layer 9808 is formed.

  A capacitor semiconductor film 9813 overlaps the capacitor wiring 9809 with a gate insulating film interposed therebetween. The capacitor semiconductor film 9813 is connected to the power supply line 9811. A portion where the capacitor semiconductor film 9813, the gate insulating film, and the capacitor wiring 9809 overlap functions as a capacitor for holding the gate voltage of the driving TFT 9802. Further, the capacitor wiring 9809 and the power supply line 9811 overlap with each other with an interlayer insulating film (not shown) interposed therebetween. A portion where the capacitor wiring 9809, the interlayer insulating film, and the power supply line 9811 overlap can function as a capacitor for holding the gate voltage of the driving TFT 9802.

  In this specification, connection means an electrical connection unless otherwise specified.

  The carrier movement directions of the channel formation regions of the active layer 9803 of the switching TFT 9801 and the active layer 9808 of the driving TFT 9802 are all aligned with the scanning direction of the laser beam indicated by the arrow.

  The number of channel formation regions included in the active layer 9808 of the driving TFT 9802 is desirably larger than the number of channel formation regions included in the active layer 9803 of the switching TFT 9801. This is because the driving TFT 9802 requires a larger current capability than the switching TFT 9801, and the on-current can be increased as the channel formation region increases.

  Note that although the structure of the TFT substrate used in the light-emitting device is described in this embodiment, a liquid crystal display device can also be manufactured using the manufacturing process of this embodiment.

  This embodiment can be implemented in combination with Embodiments 6 to 10.

  Since the TFT of the semiconductor device of the present invention has excellent crystallinity in the channel formation region, it is usually a circuit formed by an element using single crystal silicon, for example, a CPU using LSI, a memory element of various logic circuits (E.g., SRAM), a counter circuit, a divider circuit logic, and the like can be formed.

  The minimum dimension of the VLSI approaches the sub-micron region, and partial element three-dimensionalization is required to achieve higher integration. In this embodiment, a structure of a semiconductor device of the present invention having a stack structure will be described.

  FIG. 46 shows a cross-sectional view of the semiconductor device of this example. A first insulating film 9701 is formed over the substrate 9700. A first TFT 9702 is formed over the first insulating film 9701. Note that the channel width of the channel formation region of the first TFT 9702 is approximately 1 to 2 microns.

  A first interlayer insulating film 9703 is formed so as to cover the first TFT 9702, and is electrically connected to the first connection wiring 9705 and the first TFT 9702 over the first interlayer insulating film 9703. Wiring 9704 is formed.

  A second interlayer insulating film 9706 is formed so as to cover the wiring 9704 and the first connection wiring 9705. The second interlayer insulating film 9706 is formed of an inorganic insulating film, and a substance that absorbs laser light irradiated in a later step, such as a colored pigment or carbon, is applied to silicon oxide, silicon oxynitride, or the like. Use a mixture of mixed materials.

  Then, if the upper surface of the second interlayer insulating film 9706 is polished by a chemical mechanical polishing method (CMP method), the second insulating film to be formed later is planarized, and the second insulating film When the semiconductor film formed over the film is crystallized with laser light, the crystallinity can be further increased.

  A second insulating film 9708 is formed over the second interlayer insulating film 9706. A second TFT 9708 is formed over the second insulating film 9707. Note that the channel width of the channel formation region of the second insulating film 9707 is approximately 1 to 2 microns.

  A third interlayer insulating film 9709 is formed so as to cover the second TFT 9708, and is electrically connected to the second connection wiring 9711 and the second TFT 9708 on the third interlayer insulating film 9709. Wiring 9710 is formed. Note that a buried wiring (plug) 9712 is formed between the first connection wiring 9705 and the second connection wiring 9711 by a damascene process or the like.

  A fourth interlayer insulating film 9713 is formed so as to cover the wiring 9710 and the second connection wiring 9711.

  In this embodiment, the first TFT 9702 and the second TFT 9708 have a so-called stack structure in which they can be stacked with an interlayer insulating film interposed therebetween. FIG. 46A illustrates a semiconductor device having a two-layer stack structure; however, the semiconductor device may have a stack structure of three or more layers. In that case, in order to prevent the element formed in the lower layer from being irradiated with laser light, an inorganic insulating film that absorbs the laser light such as the second interlayer insulating film 9706 is provided between the layers. .

  Such a three-dimensional semiconductor device can be highly integrated, and the wiring that electrically connects each element can be shortened, so that signal delay due to the capacitance of the wiring can be prevented and higher speed can be achieved. Operation becomes possible.

  Note that the TFT using the present invention is the 4th New Functional Device Technology Symposium Proceedings, July 1985, p205. It can also be used for CAM and RAM coexisting chips described in 1. FIG. 46B shows a model in which a content addressable memory (CAM) in which a processor corresponding to a memory (RAM) is arranged and a RAM are coexisting. The first layer is a layer in which a word processing system circuit is formed, the second layer is a layer in which a processor corresponding to the third layer RAM is formed by various logic circuits, and the third layer is a RAM. A layer in which cells are formed. The second layer processor and the third layer RAM cell form an associative memory (CAM). Further, the fourth layer is a data RAM (data RAM), which coexists with an associative memory formed by the second and third layers.

  As described above, the present invention can be applied to various three-dimensional semiconductor devices.

  This embodiment can be implemented in combination with Embodiments 6-11.

  A semiconductor device including a TFT manufactured using the present invention can be applied to various electronic devices. Examples thereof include portable information terminals (electronic notebooks, mobile computers, mobile phones, etc.), video cameras, digital cameras, personal computers, television receivers, mobile phones, projection display devices, and the like. Specific examples of these electronic devices are shown in FIGS.

  FIG. 45A illustrates a display device, which includes a housing 2001, a support base 2002, a display portion 2003, a speaker portion 2004, a video input terminal 2005, and the like. By using the semiconductor device of the present invention for the display portion 2003, the display device of the present invention is completed. Since the light-emitting device is a self-luminous type, a backlight is not necessary and a display portion thinner than a liquid crystal display can be obtained. The display devices include all information display devices for personal computers, TV broadcast reception, advertisement display, and the like.

  FIG. 45B shows a digital still camera, which includes a main body 2101, a display portion 2102, an image receiving portion 2103, operation keys 2104, an external connection port 2105, a shutter 2106, and the like. By using the semiconductor device of the present invention for the display portion 2102, the digital still camera of the present invention is completed.

FIG. 45C shows a laptop personal computer, which includes a main body 2201, a housing 2202, a display portion 2203, a keyboard 2204, an external connection port 2205, a pointing mouse 2206, and the like. By using the semiconductor device of the present invention for the display portion 2203, the notebook personal computer of the present invention is completed.

  FIG. 45D shows a mobile computer, which includes a main body 2301, a display portion 2302, a switch 2303, operation keys 2304, an infrared port 2305, and the like. By using the semiconductor device of the present invention for the display portion 2302, the mobile computer of the present invention is completed.

  FIG. 45E shows a portable image reproducing device (specifically, a DVD reproducing device) provided with a recording medium, which includes a main body 2401, a housing 2402, a display portion A2403, a display portion B2404, and a recording medium (DVD or the like). A reading unit 2405, operation keys 2406, a speaker unit 2407, and the like are included. A display portion A2403 mainly displays image information, and a display portion B2404 mainly displays character information. Note that an image reproducing device provided with a recording medium includes a home game machine and the like. By using the semiconductor device of the present invention for the display portions A, B 2403 and 2404, the image reproducing device of the present invention is completed.

  FIG. 45F shows a goggle type display (head mounted display), which includes a main body 2501, a display portion 2502, and an arm portion 2503. By using the semiconductor device of the present invention for the display portion 2502, the goggle type display of the present invention is completed.

  FIG. 45G illustrates a video camera, which includes a main body 2601, a display portion 2602, a housing 2603, an external connection port 2604, a remote control reception portion 2605, an image receiving portion 2606, a battery 2607, an audio input portion 2608, operation keys 2609, and an eyepiece. Part 2610 and the like. By using the semiconductor device of the present invention for the display portion 2602, the video camera of the present invention is completed.

Here, FIG. 45H shows a cellular phone, which includes a main body 2701, a housing 2702, a display portion 2703, an audio input portion 2704, an audio output portion 2705, operation keys 2706, an external connection port 2707, an antenna 2708, and the like. Note that the display portion 2703 can suppress current consumption of the mobile phone by displaying white characters on a black background. By using the semiconductor device of the present invention for the display portion 2703, the cellular phone of the present invention is completed.

  As described above, the applicable range of the present invention is so wide that it can be used for electronic devices in various fields. In addition, this embodiment can be implemented in combination with any of the configurations shown in Embodiments 6 to 12.

  The multi-channel TFT included in the semiconductor device of the present invention has more variations in S value, mobility, threshold value, etc. than single-channel TFT and multi-channel TFT formed using a semiconductor film crystallized on a flat insulating film. Is suppressed.

  FIG. 47A shows a frequency distribution of S values of the n-type multichannel TFT of the present invention. The multi-channel TFT of the present invention uses a semiconductor film crystallized by laser light irradiation on an insulating film having unevenness. The widths of the protrusions and recesses of the insulating film are 1.25 μm and 1.50 μm, respectively, the TFT channel length is 8 μm, and the total channel width is 12 μm.

  For comparison, FIG. 47B shows a frequency distribution of S values of an n-type single channel TFT crystallized on a flat insulating film. The channel length of the TFT is 8 μm and the channel width is 8 μm. FIG. 47C shows a frequency distribution of S values of an n-type multichannel TFT crystallized on a flat insulating film. The channel length of the TFT is 8 μm, the total channel width is 12 μm, the width of each channel is 2 μm, and the interval between channels is 2 μm.

  In FIG. 47B, the standard deviation is σ = 15.8 mV / dec. 47C, the standard deviation is σ = 19.9 mV / dec. In contrast, in FIG. 47A, the standard deviation is σ = 8.1 mV / dec. It is smaller than the previous two. Therefore, it can be seen that the n-type multi-channel TFT of the present invention shown in FIG.

  Note that the channel width of the TFT in FIG. 47B is shorter than the total channel width of the TFT in FIG. In addition, the TFT of FIG. 47C has a longer width of each channel and an interval between the channels than the TFT of FIG. However, even if these conditions are taken into consideration, the standard deviation of FIG. 47A is considered to be remarkably smaller than that of FIGS. 47B and 47C. Therefore, the n-channel TFT of the present invention is considered. Is presumed to have an effect of suppressing the S value.

  Next, FIG. 48A shows a frequency distribution of threshold values of the n-type multi-channel TFT of the present invention. The structure of the TFT in FIG. 48A is the same as that in FIG. For comparison, FIG. 48B shows a frequency distribution of threshold values of an n-type single channel TFT crystallized on a flat insulating film. The structure of the TFT in FIG. 48B is the same as that in FIG. FIG. 48C shows a frequency distribution of threshold values of an n-type multichannel TFT crystallized on a flat insulating film. The structure of the TFT in FIG. 48C is the same as that in FIG.

  In FIG. 48B, the standard deviation is σ = 126 mV, and in FIG. 48C, the standard deviation is σ = 153 mV, whereas in FIG. 48A, the standard deviation is σ = 80 mV. Smaller than that. Therefore, it can be seen that the variation in threshold value is suppressed in the n-type multi-channel TFT of the present invention shown in FIG.

  Note that the channel width of the TFT in FIG. 48B is shorter than the total channel width of the TFT in FIG. In addition, the TFT of FIG. 48C has a longer width of each channel and an interval between channels than the TFT of FIG. However, even if these conditions are taken into consideration, the standard deviation of FIG. 48A is considered to be remarkably smaller than those of FIGS. 48B and 48C. Therefore, the n-channel TFT of the present invention is considered. Is presumed to have an effect that the threshold is suppressed.

  Next, FIG. 49A shows a frequency distribution of mobility of the n-type multi-channel TFT of the present invention. The structure of the TFT in FIG. 49A is the same as that in FIG. For comparison, FIG. 49B shows a frequency distribution of mobility of an n-type single channel TFT crystallized on a flat insulating film. The structure of the TFT in FIG. 49B is the same as that in FIG. FIG. 49C shows a frequency distribution of mobility of an n-type multichannel TFT crystallized on a flat insulating film. The structure of the TFT in FIG. 49C is the same as that in FIG.

  In FIG. 49B, the standard deviation is σ = 7.9%, and in FIG. 49C, the standard deviation is σ = 9.2%, whereas in FIG. 49A, the standard deviation is σ = 5.2%. It is smaller than the previous two. Therefore, it can be seen that the n-type multi-channel TFT of the present invention shown in FIG. In FIG. 49A, the mobility is calculated using the channel width design value, so the actual mobility is considered to be about 20% lower.

The channel width of the TFT in FIG. 49B is shorter than the total channel width of the TFT in FIG. In addition, the TFT of FIG. 49C has a longer width of each channel and an interval between the channels than the TFT of FIG.
However, even if these conditions are taken into consideration, the standard deviation in FIG. 49A is considered to be remarkably smaller than those in FIGS. 49B and 49C. Therefore, the n-channel TFT of the present invention is considered. Is presumed to have the effect of reducing mobility.

  Next, FIG. 50A shows a frequency distribution of threshold values of the p-type multichannel TFT of the present invention. The structure of the TFT in FIG. 50A is the same as that in FIG. 47A except that the polarity is different. For comparison, FIG. 50B shows a frequency distribution of threshold values of a p-type single channel TFT crystallized on a flat insulating film. The structure of the TFT in FIG. 50B is the same as that in FIG. 47B except that the polarities are different. FIG. 50C shows a frequency distribution of threshold values of a p-type multichannel TFT crystallized on a flat insulating film. The structure of the TFT in FIG. 50C is the same as that in FIG. 47C except that the polarity is different.

  In FIG. 50B, the standard deviation is σ = 218 mV, and in FIG. 50C, the standard deviation is σ = 144 mV, whereas in FIG. 50A, the standard deviation is σ = 77 mV. Smaller than that. Therefore, it can be seen that the p-type multi-channel TFT of the present invention shown in FIG.

  Note that the channel width of the TFT in FIG. 50B is shorter than the total channel width of the TFT in FIG. In addition, the TFT of FIG. 50C has a longer width of each channel and an interval between the channels than the TFT of FIG. However, even if these conditions are taken into consideration, the standard deviation of FIG. 50A is considered to be remarkably smaller than that of FIGS. 50B and 50C. Therefore, the p-channel TFT of the present invention is considered. Is presumed to have an effect that the threshold is suppressed.

  Next, FIG. 51A shows the frequency distribution of the mobility of the p-type multi-channel TFT of the present invention. The structure of the TFT in FIG. 51A is the same as that in FIG. 47A except that the polarity is different. For comparison, FIG. 51B shows a frequency distribution of mobility of a p-type single channel TFT crystallized on a flat insulating film. The structure of the TFT in FIG. 51B is the same as that in FIG. 47B except that the polarity is different. FIG. 51C shows a frequency distribution of mobility of a p-type multichannel TFT crystallized on a flat insulating film. The structure of the TFT in FIG. 51C is the same as that in FIG. 47B except that the polarity is different.

  In FIG. 51B, the standard deviation is σ = 7.6%, and in FIG. 51C, the standard deviation is σ = 5.9%, whereas in FIG. 51A, the standard deviation is σ = 4.6%. It is smaller than the previous two. Therefore, it can be seen that the p-type multi-channel TFT of the present invention shown in FIG. In FIG. 51A, since the mobility is calculated using the design value of the channel width, the actual mobility is considered to be about 20% lower.

  Note that the channel width of the TFT in FIG. 51B is shorter than the total channel width of the TFT in FIG. In addition, the TFT in FIG. 51C has a longer width of each channel and a gap between the channels than the TFT in FIG. However, even if these conditions are taken into consideration, the standard deviation of FIG. 51A is considered to be remarkably smaller than that of FIGS. 51B and 51C. Therefore, the p-channel TFT of the present invention is considered. Is presumed to have the effect of reducing mobility.

  As shown in FIGS. 47 to 51, the multi-channel TFT of the present invention has an effect of suppressing variation in characteristics. The multi-channel TFT of the present invention includes various crystal orientations because the crystal orientation of each channel is more easily rotated than the single-channel TFT and multi-channel TFT crystallized on a flat insulating film. Therefore, it is considered that the variation in characteristics due to the crystal orientation is easily averaged.

The perspective view explaining the crystallization method in this invention. The perspective view explaining the crystallization method in this invention. The perspective view explaining the crystallization method in this invention. The perspective view explaining the crystallization method in this invention. The longitudinal cross-sectional view explaining the detail of the relationship between the shape of the opening part in crystallization, and the form of a crystalline semiconductor film. 1 is a layout diagram illustrating one embodiment of a laser irradiation apparatus applied to the present invention. 8A and 8B are a top view and a vertical cross-sectional view illustrating a manufacturing process of a TFT manufactured according to the present invention. 8A and 8B are a top view and a vertical cross-sectional view illustrating a manufacturing process of a TFT manufactured according to the present invention. 8A and 8B are a top view and a vertical cross-sectional view illustrating a manufacturing process of a TFT manufactured according to the present invention. 8A and 8B are a top view and a vertical cross-sectional view illustrating a manufacturing process of a TFT manufactured according to the present invention. 8A and 8B are a top view and a vertical cross-sectional view illustrating a manufacturing process of a TFT manufactured according to the present invention. The top view and longitudinal cross-sectional view explaining an example of TFT produced by this invention. The top view and longitudinal cross-sectional view explaining an example of TFT produced by this invention. The top view and longitudinal cross-sectional view explaining an example of TFT produced by this invention. The external view of a display panel. FIG. 6 is a top view illustrating a structure of a pixel portion of a display panel manufactured using the present invention. Sectional drawing which shows the structure used for the simulation of thermal analysis. The graph which shows the result of the simulation of thermal analysis. A cross-sectional image of a TEM after a semiconductor film formed on a base film having a convex portion is irradiated with laser light to be crystallized, and a schematic diagram thereof. The figure which shows the flow of crystallization of the semiconductor film of this invention. 4A and 4B illustrate a manufacturing process of a TFT of the present invention. 4A and 4B illustrate a manufacturing process of a TFT of the present invention. 4A and 4B illustrate a manufacturing process of a TFT of the present invention. 4A and 4B illustrate a manufacturing process of a TFT of the present invention. 4A and 4B illustrate a manufacturing process of a TFT of the present invention. Sectional drawing of TFT of this invention. Sectional drawing of TFT of this invention. 4A and 4B illustrate a manufacturing process of a TFT of the present invention. 4A and 4B illustrate a manufacturing process of a TFT of the present invention. FIG. 6 is a top view of a plurality of TFTs formed on a base film. The figure which shows the flow of crystallization of the semiconductor film of this invention using a catalyst element. The figure of a laser irradiation apparatus. The figure of a laser irradiation apparatus. 10A and 10B illustrate a method for manufacturing a base film having a convex portion. 10A and 10B illustrate a method for manufacturing a base film having a convex portion. The figure which shows distribution of the energy density of a laser beam. The figure which shows distribution of the energy density of a laser beam. The figure which shows distribution of the energy density of a laser beam. FIG. The figure which shows distribution of the energy density in the central-axis direction of the superimposed laser beam. The figure which shows the relationship between the distance between the centers of a laser beam, and an energy difference. The figure which shows distribution of the output energy in the center axis direction of a laser beam. FIG. 6 illustrates a structure of a light-emitting device that is an example of a semiconductor device of the present invention. 3A and 3B each illustrate a structure of a pixel of a light-emitting device that is an example of a semiconductor device of the present invention. FIG. 16 is a diagram of an electronic device using the semiconductor device of the invention. Sectional drawing of TFT which has a stack structure, and an example of a structure of a semiconductor device using the same. The figure which shows the frequency distribution of S value. The figure which shows the frequency distribution of a threshold value. The figure which shows the frequency distribution of a mobility. The figure which shows the frequency distribution of a threshold value. The figure which shows the frequency distribution of a mobility.

Claims (12)

A substrate including a pixel portion, a plurality of input terminals, and a wiring or a wiring group connected to the plurality of input terminals;
A semiconductor device having a first FPC and a second FPC fixed to the plurality of input terminals. A substrate including a pixel portion, a plurality of input terminals, and a wiring or a wiring group connected to the plurality of input terminals;
A first FPC and a second FPC fixed to the plurality of input terminals;
A semiconductor device comprising: a chip mounted on the substrate between the first FPC and the second FPC.   3. The semiconductor device according to claim 2, wherein the chip is mounted on the substrate by a COG method. A substrate provided with a pixel portion, a gate signal side drive circuit, a data signal side drive circuit, a plurality of input terminals, and a wiring or a wiring group connected to the plurality of input terminals;
A first FPC and a second FPC fixed to the plurality of input terminals;
A semiconductor device comprising: a seal pattern formed outside the gate signal side driving circuit.   5. The semiconductor device according to claim 4, wherein the seal pattern partially overlaps the gate signal side driving circuit. A substrate provided with a pixel portion, a gate signal side drive circuit, a data signal side drive circuit, a plurality of input terminals, and a wiring or a wiring group connected to the plurality of input terminals;
A first FPC and a second FPC fixed to the plurality of input terminals;
And a seal pattern formed so as to partially overlap the outside of the data signal side driving circuit. A substrate provided with a pixel portion, a gate signal side drive circuit, a data signal side drive circuit, a plurality of input terminals, and a wiring or a wiring group connected to the plurality of input terminals;
A first FPC and a second FPC fixed to the plurality of input terminals;
A semiconductor device comprising: the gate signal side driving circuit, the data signal side driving circuit, and a seal pattern formed to partially overlap the outside of the wiring or the wiring group.   8. The semiconductor device according to claim 4, wherein the wiring or the wiring group connects the gate signal side driving circuit or the data signal side driving circuit and the plurality of input terminals. .   9. The semiconductor device according to claim 4, further comprising a chip mounted on the substrate between the first FPC and the second FPC.   10. The semiconductor device according to claim 9, wherein the chip is mounted on the substrate by a COG method.   11. An electronic apparatus, wherein the semiconductor device according to claim 1 is used for a display portion. 12. The electronic device according to claim 11, wherein the electronic device is any one of a mobile phone, a portable information terminal, a video camera, a digital camera, and a television receiver.

Images