JP2006135348A - Semiconductor device and manufacturing method thereof - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To obtain a thin film transistor (TFT) in which a leak current (OFF current) is low when reverse bias voltage is applied especially to a gate electrode. <P>SOLUTION: A semiconductor device includes the thin film transistor comprising an active layer formed on an insulation surface and consisting of non-monocrystal silicon, a gate insulating film formed in contact with the active layer, and a gate electrode formed in contact with the gate insulating film. The active layer comprises a channel formation area, a source area in which the concentration of oxygen, carbon and nitrogen is higher than that of the channel formation area and having n-type or p-type impurity, and a drain area. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a thin film transistor (TFT) having a non-single crystal silicon film provided on an insulating substrate such as glass or an insulating film formed on various substrates, or a thin film integrated circuit to which the thin film integrated circuit is applied. The present invention relates to a thin film integrated circuit for a liquid crystal display device (liquid crystal display) and a manufacturing method thereof.

  In recent years, semiconductor devices having TFTs on an insulating substrate such as glass have been developed, for example, active liquid crystal display devices using TFTs for driving pixels, image sensors, three-dimensional integrated circuits, and the like.

  A thin film silicon semiconductor is generally used for TFTs used in these devices. Thin film silicon semiconductors are roughly classified into two types: those made of amorphous silicon semiconductor (a-Si) and those made of crystalline silicon semiconductor. Amorphous silicon semiconductors are most commonly used because they have a low production temperature, can be produced relatively easily by a vapor phase method, and are highly mass-productive. However, physical properties such as conductivity have crystallinity. Since it is inferior to a silicon semiconductor, the establishment of a method for manufacturing a TFT made of a crystalline silicon semiconductor has been strongly demanded in order to obtain higher speed characteristics in the future. As silicon semiconductors having crystallinity, polycrystalline silicon, microcrystalline silicon, amorphous silicon containing a crystalline component, semi-amorphous silicon having a state between crystalline and amorphous are known. .

  Such amorphous or polycrystalline silicon (collectively referred to as non-single crystal) silicon has a much larger grain boundary effect than single crystal materials used in single crystal semiconductor integrated circuits. It was. A typical example was a source / drain leakage current. In particular, when a reverse bias voltage (that is, negative for an N-channel transistor and positive for a P-channel transistor) is applied to the gate electrode, leakage current (also referred to as off-current) is due to the effect of the grain boundary. This deteriorates the operating characteristics of the transistor and has been demanded for improvement.

  Such an off-current is generated because an electric field fluctuates rapidly at the boundary between the source / drain impurity region (N-type or P-type) and the channel formation region (substantially intrinsic). Although not a problem, in the non-single crystal, carriers hop through the grain boundary from the conduction band (valence band) of the impurity region to the valence band (conduction band) of the channel formation region. Therefore, as in the case of a single crystal MOS device, an attempt has been made to solve the problem by providing an offset region in order to relax the electric field, or by using a low concentration drain (LDD) structure that lowers the impurity concentration of the impurity region. .

  FIG. 2A shows a conceptual diagram of the structure of a conventional offset gate type TFT. The active layer is roughly divided into five regions. The first region is an impurity region (source / drain) having a high impurity concentration and corresponds to the regions 13 and 17. The second region is a region called an offset region or an LDD region, and corresponds to the regions 14 and 16, and this region is substantially the same as the source / drain in a region that suppresses generation of a parasitic channel. Although it is a conductive type, it has a high resistance, and there is no gate electrode thereover. The third region is a region called a channel formation region and corresponds to the region 15, and the conductivity type can be changed through the gate insulating film 12 by the influence of the gate electrode 11 to control the ease of carrier flow.

FIG. 2B shows the energy band in the vicinity of the gate insulating film of the active layer when no voltage is applied to the gate electrode and the source / drain voltage is sufficiently low in the N-channel TFT. Here, E _F is a Fermi surface, E _I and E _N are energy band gaps of the channel formation region and the impurity region, and usually E _I = E _N. In addition, the band gap of the offset region is the same as E _I. When a reverse bias (negative) voltage is applied to the gate electrode while keeping the source-drain voltage unchanged, the band diagram changes as shown in FIG. Here, the electric field portion immediately below the gate electrode of the active layer changes by E _G.

  It should be noted that the presence of the offset regions 14 and 16 smoothly changes the electric field between the impurity region and the channel formation region, and has an effect of suppressing a leakage current therebetween. However, when a large forward bias (positive) voltage is applied between the source / drain in this state, the band of the drain region 17 is lower than that indicated by the solid line in the figure, and therefore, between the channel forming region 15 and the drain region 17. This electric field becomes steeper in spite of the presence of the offset region 16, and a leak current passing through the grain boundary is generated.

Even in actual measurement, the off-current is not so noticeable while the source-drain voltage V _D is small, but the off-current increases as V _D increases, and the reverse bias voltage (in the case of an N-channel TFT) This is proved by the experimental fact that the off-current increases as the negative voltage increases. Therefore, in order to reduce this off-current (especially in a region where V _D is large), the electric field between the drain region and the channel formation region changes smoothly even when V _D is large. A TFT having such a structure must be manufactured.

In particular, the reduction of the off-state current is required when the active layer has a small amount of a metal element for promoting crystallization. Examples of metal elements that promote crystallization include Ni, Fe, Co, Ni, Ru, Rh, Pd, Os, Ir, Pt, Sc, Ti, V, Cr, Mn, Cu, Zn, Au, and Ag. By adding these elements to the silicon film at a concentration of 1 × 10 ¹⁸ to 2 × 10 ²¹ atoms cm ⁻³ , the crystallization temperature of amorphous silicon is lowered and the crystallization time is reduced. It can be shortened. However, in silicon films to which these elements are added, intermediate levels may be generated in the band gap by these elements, and this level has the same effect as the grain boundary, which increases the off current. It became.

A first method for solving such a problem is to increase the band gap of the semiconductor. That is, even in the conventional offset gate type TFT as shown in FIG. 2, if the band gap of the semiconductor is increased, the band steepness between the channel formation region and the drain region is the same, but between the conduction band and the valence band. Therefore, the effect of the offset region is increased, and carrier hopping from the valence band (conduction band) of the channel formation region to the conduction band (valence band) of the drain region is suppressed. In this way, increasing the band gap of a silicon semiconductor can be realized by adding an appropriate amount of an element having an effect of increasing the energy band width of carbon, nitrogen, oxygen or the like. The larger the amount added, the band gap increases, but excessive addition degrades the semiconductor characteristics. According to the inventor's research, it is clear that the total concentration of these elements is 1 × 10 ¹⁹ to 2 × 10 ²¹ cm ⁻³ , preferably 5 × 10 ^{19 to} 7 × 10 ²⁰ cm ^−3. became. Note that the measurement of the concentration of the impurity element means the lowest value in the SIMS (secondary ion mass spectrometry) method.

The disadvantage of the first method is that oxygen, nitrogen, carbon, etc. are also added to the channel region, so that the current (on-current) when a forward bias (positive) voltage is applied to the gate electrode is reduced. It is to end. In order to solve this problem, as shown in FIG. 1A, the impurity region 3 can be substantially suppressed between the impurity regions 3 and 7 and the channel formation region 5 within a range that substantially suppresses the generation of intrinsic or parasitic channels. , 7 and regions 4 and 6 having the same conductivity type and a large band gap as those of the first and second layers (second method). A band diagram in the vicinity of the gate electrode of the N-channel TFT having such a region is shown in FIG. 1B when no voltage is applied to the gate electrode. That is, E _I <E _O and generally E _I <E _N. That _is, the relationship of _{E I <E O ≦ E N} , or _N I _{<N O} ≦ relationship _{N N} (where, N is the added carbon, oxygen, concentration of nitrogen) with a.

A band diagram when a negative voltage is applied to the gate electrode is shown in FIG. As is apparent from the figure, the presence of the regions 4 and 6 significantly hinders carrier movement between the conduction band (valence band) of the impurity region and the valence band (conduction band) of the channel formation region. As a result, as shown in FIG. 1D, the off-current decreases, and even when V _D is large, the off-current is more stable than the conventional offset gate type TFT.

Such regions 4 and 6 can be obtained by adding an appropriate amount of carbon, nitrogen, oxygen or the like as in the first method. According to the study of the present inventor, the total concentration of these elements is 1 × 10 ¹⁹ to 2 × 10 ²¹ atoms cm ⁻³ , preferably 5 × 10 ^{19 to} 7 × 10 ²⁰ atoms cm ⁻³ . Generally, when adding these elements to the regions 4 and 6, at the same time common to add the same amount to the impurity regions 3,7, At that time, in FIG. 2 (B), E _O = E _N (or N _O = N _N ).

In the first and second methods, an ion doping method or an ion implantation method may be used to add oxygen, carbon, nitrogen, etc., but this damages the crystallinity of the silicon film. Thereafter, thermal annealing, laser annealing, lamp annealing (RTA) or the like is crystallized, and these additives are uniformly bonded to silicon and neutralize the grain boundaries. Alternatively, it is desirable that the N-type impurity is sufficiently activated. In particular, regarding the first method, an appropriate amount of oxygen, nitrogen, carbon, or the like may be mixed during the formation of the silicon film. In the first method, the concentrations of oxygen, nitrogen, and carbon do not change between the source / drain and the channel formation region, and therefore N _I = N _O = N _N. An example of a method for manufacturing a TFT will be described below, and a TFT based on the first and second methods will be described.

  As is clear from the examples, the present invention can reduce the off-current of the TFT. As shown in the examples, the TFT containing a metal element (nickel or the like) for promoting crystallization in the active layer was particularly effective in reducing the off-current. Further, as shown in Embodiment 4, in the present invention, the silicide layer is formed in close contact with the source / drain, thereby reducing the substantial sheet resistance of the source / drain, and thus the source / drain dose. By reducing the amount, the effect of the present invention could be further enhanced.

  The present invention can be applied not only to N-channel TFTs but also to P-channel TFTs. However, in the present invention, when elements such as oxygen, nitrogen, and carbon are introduced into the channel formation region as in the first method (Example 1 and Example 2), the threshold value due to these elements is set. Voltage fluctuations must be taken into account. For example, in the case of oxygen, 1 to 10% of the chlorinated substances in silicon serve as donors, so that the N-type is weak. Therefore, the threshold value shifts in the positive direction in both the P-channel TFT and the N-channel TFT.

  In the present invention, particularly in the second method shown in Embodiments 3 and 4, elements such as oxygen, carbon, and nitrogen are introduced into the boundary portion between the impurity region and the channel formation region. Conventionally, this boundary has been a problem of large distortion due to impurity doping, laser irradiation, etc., but the presence of these elements alleviates such lattice distortion, resulting in the effect of reducing off-current. Have

  The present invention can obtain a greater effect not only in a semiconductor circuit in which the same type of TFT is formed on the same substrate but also in a semiconductor integrated circuit in which different types of TFT are formed on the same substrate. it can. For example, a part of the circuit is a conventional self-aligned TFT or offset gate type TFT without offset, and the other part of the circuit uses the TFT of the present invention.

  For example, in an active matrix substrate used for a liquid crystal display, in a monolithic thin film integrated circuit in which an active matrix circuit and a peripheral circuit for driving the same are formed on the same substrate, the TFT used for the active matrix circuit is: Since it is necessary to hold the charge accumulated in the pixel for a long time, it is preferable that the off-state current is small, and a TFT manufactured using the present invention is suitable.

  On the other hand, TFTs with large on-currents are suitable for TFTs used for peripheral circuits because of the high operating frequency required. However, the TFT according to the present invention tends to be slightly smaller than the conventional one in terms of on-current. This tendency is particularly strong in the TFTs shown in Examples 1 and 2. Therefore, it is preferable to use a conventional self-aligned type or offset gate type TFT for the peripheral circuit.

  Thus, by making the active matrix circuit and the TFT of the peripheral circuit different, the characteristics of the entire circuit can be greatly improved. As described above, the present invention is industrially useful.

  FIG. 3 shows a cross-sectional view of a manufacturing process of a TFT according to this embodiment. First, a base film 102 of silicon oxide having a thickness of 200 nm was formed on a substrate (Corning 7059) 101 by a sputtering method. The substrate is annealed at a temperature higher than the strain temperature before or after the formation of the base film, and then slowly cooled to below the strain temperature at 0.1 to 1.0 ° C./min. There is little shrinkage | contraction of a board | substrate in the accompanying process, and mask alignment becomes easy. In the Corning 7059 substrate, after annealing at 620 to 660 ° C. for 1 to 4 hours, 0.01 to 1.0 ° C./min, preferably 0.03 to 0.3 ° C./min, and then slowly cooled to 400 to 500 It is good to take it out when the temperature drops to ℃.

  Then, an intrinsic (I-type) amorphous silicon film 103 having a thickness of 20 to 200 nm, preferably 30 to 150 nm, for example, 120 nm was formed by plasma CVD. Further, a silicon oxide film 104 having a thickness of 10 to 80 nm, for example, 20 nm was deposited thereon by plasma CVD. This becomes a protective film in the following thermal annealing process, and prevents the film surface from being rough.

Furthermore, oxygen ions were implanted at a dose of 1 × 10 ^{14 to} 3 × 10 ¹⁶ cm ⁻² , for example, 2 × 10 ¹⁵ cm ⁻² by ion implantation. Implantation was performed by selecting energy so that a peak would appear at the boundary between the amorphous silicon film 103 and the base film 102. As a result, it was confirmed by the secondary ion mass spectrometry (SIMS) method that oxygen was introduced into the amorphous silicon film 103 at a concentration of 5 × 10 ¹⁹ to 2 × 10 ²¹ cm ⁻³ (depending on the depth). confirmed. This addition of oxygen atoms may be performed simultaneously with the formation of the semiconductor film. (Fig. 3 (A))

  Next, the silicon film was crystallized by thermal annealing at 600 ° C. for 48 hours in a nitrogen atmosphere (atmospheric pressure). Further, the silicon oxide film 104 was removed, and the silicon film was patterned to form an island-like active layer 105 of a TFT made of an intrinsic or substantially intrinsic polycrystalline silicon film. The size of the active layer 105 is determined in consideration of the channel length and channel width of the TFT. The smaller one was 50 μm × 20 μm, and the larger one was 100 μm × 1000 μm. Many such active layers were formed on the substrate.

Then, a silicon oxide film 106 having a thickness of 100 nm was formed as a gate insulating film by plasma CVD. As a CVD source gas, TEOS (tetraethoxysilane, Si (OC ₂ H ₅ ) ₄ ) and oxygen were used, and the substrate temperature during film formation was 300 to 550 ° C., for example, 400 ° C. This may be an oxide film formed by thermal oxidation.

  Subsequently, an aluminum film (including 0.01 to 0.25% scandium or other group IIIa element (rare earth element)) having a thickness of 300 to 800 nm, for example, 600 nm was formed by a sputtering method. The gate electrode 107 was formed by patterning the aluminum film. In addition to aluminum, a semiconductor to which a metal such as tantalum or titanium or phosphorus is added may be used as the gate electrode material. (Fig. 3 (B))

  Further, the surface of the aluminum electrode was anodized to form an oxide layer 108 on the surface. This anodization was performed in an ethylene glycol solution containing 1 to 5% tartaric acid. The thickness of the obtained oxide layer 108 was 200 nm. Note that the oxide 108 has a thickness for forming an offset gate region in a subsequent ion doping step, and thus the length of the offset gate region can be determined in the anodic oxidation step. (Figure 3 (C))

Next, by an ion doping method (also referred to as a plasma doping method), the gate electrode portion (that is, the gate electrode 107 and the surrounding oxide layer 108) is used as a mask to add impurities that impart N conductivity type in a self-aligning manner to the silicon film 105. Added to. As the doping gas, phosphine (PH ₃ ) was used, and the acceleration voltage was set to 60 to 90 kV, for example, 80 kV. The dose amount was 1 × 10 ¹⁵ to 8 × 10 ¹⁵ cm ⁻² , for example, 5 × 10 ¹⁵ cm ⁻² . As a result, N-type impurity regions 109 and 110 to be the source / drain were formed.

Thereafter, annealing was performed by laser light irradiation to activate the doped impurities. As the laser light, a KrF excimer laser (wavelength 248 nm, pulse width 20 nsec) was used, but other lasers may be used. The laser light was irradiated under the conditions of an energy density of 200 to 400 mJ / cm ² , for example, 250 mJ / cm ^2, and ² to 10 shots, for example, 2 shots were irradiated at one place. The effect may be increased by heating the substrate to about 200 to 450 ° C. during the laser light irradiation. (Fig. 3 (D))

Further, this step may be a method by lamp annealing (rapid thermal annealing, RTA) using visible or near infrared light. Visible and near infrared rays are easily absorbed by crystallized silicon, or amorphous silicon to which phosphorus or boron is added at 10 ¹⁹ to 10 ²¹ cm ⁻³ , and effective annealing comparable to thermal annealing at 1000 ° C. or higher. It can be performed. When phosphorus or boron is added, light is sufficiently absorbed even in the near infrared due to the impurity scattering. This can be fully inferred from the fact that it is black even when observed with the naked eye. On the other hand, it is difficult to be absorbed by the glass substrate, so it does not heat the glass substrate to a high temperature and requires only a short processing time, so it can be said that it is an optimal method in the process where shrinkage of the glass substrate is a problem. .

  After the impurity activation step, a silicon oxide film 111 having a thickness of 600 nm was formed as an interlayer insulator by a plasma CVD method. As the interlayer insulator, polyimide or a two-layer film of silicon oxide and polyimide may be used. Further, a contact hole was formed larger than usual, and first, titanium or nickel was formed to 10 to 50 nm on the entire surface. Further, these were irradiated with laser or strong light of visible or near infrared light to react with silicon in the source / drain regions 109 and 110 to form silicide regions 112 and 113. Thereafter, titanium or nickel that did not react with silicon was removed by etching, and TFT electrodes / wirings 114 and 115 were formed of a metal material, for example, a multilayer film of titanium nitride and aluminum. Finally, annealing was performed at 350 ° C. for 30 minutes in a hydrogen atmosphere of 1 atm to complete the TFT. (Figure 3 (E))

In this way, the off-current when the gate voltage was 0 V could be lowered from the conventional 1 × 10 ⁻¹⁰ A to 3 × 10 ⁻¹² A. Furthermore, when the drain voltage is +10 V and the gate voltage is −10 V, the leakage current (off current) is 3 to 5 × 10 ⁻¹² A, which can be reduced to about ^1/30 of the conventional one.

  An outline of the manufacturing process of this example is shown in FIG. In this example, a Corning 7059 glass substrate (thickness 1.1 mm, 300 × 400 mm) was used as the substrate 201. First, the substrate 201 is annealed at 620 to 660 ° C. for 1 to 4 hours, and then slowly cooled at 0.01 to 1 ° C./min, preferably 0.03 to 0.3 ° C./min, and lowered to 400 to 500 ° C. At that stage, the glass substrate was taken out to room temperature and the glass substrate was shrunk. Further, a base film 202 (silicon oxide) was formed on the substrate 201 to a thickness of 200 nm by plasma CVD. TEOS and oxygen were used as the source gas for CVD. Further, an amorphous silicon film 203 was formed to a thickness of 100 nm and a silicon oxide film 204 was formed to a thickness of 200 nm by LPCVD or plasma CVD. Then, holes as shown in the region 205 were formed in the silicon oxide film by a known photolithography method. Thereafter, a nickel film 206 having a thickness of 0.5 to 2 nm was formed by sputtering. The nickel film 206 does not have to be a film. In addition to nickel, any element that promotes crystallization of amorphous silicon may be used.

Furthermore, nitrogen ions were implanted at a dose of 1 × 10 ^{14 to} 3 × 10 ¹⁶ cm ⁻² , for example, 2 × 10 ¹⁵ cm ⁻² by ion implantation. The implantation was performed by selecting energy so that a peak was reached at the boundary between the amorphous silicon film 203 and the silicon oxide film 204. As a result, it was confirmed by secondary ion mass spectrometry (SIMS) that nitrogen was introduced into the amorphous silicon film 203 at a concentration of 1 × 10 ²⁰ to 1 × 10 ²¹ cm ⁻³ (depending on the depth). confirmed. (Fig. 2 (A))

  Then, the silicon film was crystallized by performing thermal annealing at 550 ° C. for 8 hours or at 600 ° C. for 4 hours. At that time, as nickel diffused through the silicon film, crystallization progressed laterally from the hole 205. The length of crystal growth was typically 20-100 μm. In FIG. 4B, in the silicon film, a region 207 is a crystallized region and a region 208 is an uncrystallized region. (Fig. 4 (B))

  After this thermal annealing step, the silicon oxide film 204 was removed, and the silicon film was patterned to form a TFT island-like active layer 209. Further, a silicon oxide gate insulating film (thickness 70 to 120 nm, typically 120 nm) 210 was formed by plasma CVD in an oxygen atmosphere using tetraethoxysilane (TEOS) as a raw material. The substrate temperature was 350 ° C. Further, aluminum (containing 0.01 to 0.2% scandium or other group IIIa element (rare earth element)) having a thickness of 600 to 800 nm, for example, 600 nm is formed by sputtering, and the aluminum film is patterned. A gate electrode was formed. Then, anodization was performed in the same manner as in Example 1 to form an anodized layer of aluminum oxide on the side and top surfaces of the gate electrode. Thus, the gate electrode portion 211 was obtained. (Fig. 4 (C))

Thereafter, phosphorus was implanted as an N-type impurity by an ion doping method to form a source region 211 and a drain region 212 in a self-aligning manner. Then, the crystallinity of the silicon film whose crystallinity deteriorated due to ion implantation was improved by irradiating with KrF laser light. At this time, the energy density of the laser beam was 250 to 300 mJ / cm ² . By this laser irradiation, the sheet resistance of the source / drain of this TFT became 1-8 kΩ / cm ² . Further, this step may be performed by visible / near infrared lamp annealing. By this annealing step, the active layer was all crystallized. (Fig. 4 (D))

  Thereafter, an interlayer insulator 213 was formed from silicon oxide or polyimide, contact holes were further formed, and electrodes 214 and 215 were formed from a chromium / aluminum multilayer film in the source / drain regions of the TFT. Finally, hydrogenation was performed by annealing in hydrogen at 200 to 400 ° C. for 1 hour. The silicon in the source / drain region and the electrode material reacted to form silicide regions 216 and 217 in the contact portion. In this way, a TFT was completed. In order to further improve the moisture resistance, a passivation film may be further formed of silicon nitride, aluminum nitride or the like on the entire surface. (Fig. 4 (E))

The TFT shown in this example is obtained by crystallization by annealing at a lower temperature and in a shorter time than that of Example 1, but nickel is 3 × 10 ^{17 to} 5 × 10 ¹⁹ cm in the active layer. ^-3 was confirmed by secondary ion mass spectrometry (SIMS). In the prior art, the leakage current increased for this reason, but in this example, nitrogen was also 1 × 10 ¹⁹ to 2 × 10 ²¹ atoms cm ⁻³ in the active layer, typically 1 × 10 ²⁰ to Due to the presence of 5 × 10 ²⁰ atoms cm ⁻³ , the leakage current was reduced, and better off characteristics were obtained.

  In this example, nitrogen introduced into the silicon film firmly terminated the dangling bonds of silicon. Oxygen did not have a significant effect, but a similar effect can be observed by introducing carbon. The dangling bonds of silicon are particularly prominent at the grain boundaries and cause leakage current, but can be suppressed by the present invention. Similarly, metal elements such as nickel present in the silicon film can be fixed by nitrogen or carbon, which is also preferable for semiconductor characteristics.

  The present embodiment will be described with reference to FIG. The substrate used was Corning 7059, and after annealing at 640 ° C. for 4 hours in advance to prevent shrinkage as in Example 2, it was gradually cooled to 450 ° C. at 0.1 ° C. and then taken out. did. First, a base film 302 was formed on a substrate 301, and an amorphous silicon film having a thickness of 30 to 80 nm, for example, 80 nm and a silicon oxide film having a thickness of 200 nm were further formed by plasma CVD. Then, holes were selectively formed in the silicon oxide film as in Example 2, and a nickel film having a thickness of 0.5 to 2 nm was formed thereon by sputtering.

  The silicon film was crystallized by annealing at 600 ° C. for 4 hours in a nitrogen atmosphere. Thereafter, the silicon oxide film was removed to expose the silicon film surface. Then, heat annealing was performed at 550 ° C. for 1 hour in an oxidizing atmosphere such as oxygen, ozone, or dinitrogen monoxide to form a thin silicon oxide film on the silicon film surface. Then, the silicon film was patterned to form an active layer 303. The silicon oxide layer on the surface of the silicon film was removed.

  Then, a thin silicon oxide film 304 was formed on the surface of the active layer by performing thermal annealing again at 600 ° C. for 1 hour in an oxidizing atmosphere. After the thermal annealing, the substrate was rapidly cooled to 450 ° C. at a rate of 2 ° C./second or more, preferably 10 ° C./second or more. This is to prevent the substrate from shrinking due to this thermal annealing step. In such an annealing furnace that cannot be rapidly cooled, the same effect can be obtained by taking the substrate out of the furnace and leaving it at room temperature. (Fig. 5 (A))

Thereafter, a silicon oxide gate insulating film 305 and an aluminum gate electrode 306 were formed in the same manner as in Example 1. Needless to say, a semiconductor to which a metal such as tantalum or titanium or phosphorus is added in addition to aluminum may be used as the gate electrode material. Then, by ion doping, oxygen ions were implanted in a self-aligned manner using the gate electrode 306 as a mask to form regions 307 and 308 with high oxygen concentration in the active layer. The dose was set to 5 × 10 ¹⁵ cm ^−2, and the energy was selected so that a peak appeared at the boundary between the active layer 303 and the gate oxide silicon oxide film 305. As a result, it was confirmed that oxygen was introduced into the regions 307 and 308 of the active layer at a concentration of 5 × 10 ^{19 to} 5 × 10 ²⁰ cm ⁻³ (depending on the depth). Confirmed by. In this case, N _I <N _O = N _N (or E _I <E _O = E _N ). (Fig. 5 (B))

Thereafter, the gate electrode was anodized in the same manner as in Example 1 to form an anodized layer 309. (FIG. 5C) Then, in the same manner as in Example 1, using the gate electrode 306 and its anodic oxide layer 309 as a mask, the active layer is doped with impurities (phosphorus) in a self-aligned manner, thereby forming impurity regions 310 and 311. Formed. The dose amount was 1 × 10 ¹³ to 1 × 10 ¹⁶ cm ⁻² , for example, 5 × 10 ¹⁵ cm ⁻² for an N ⁺ type. And this was activated by laser irradiation. (Fig. 5 (D))

Further, an interlayer insulator 312 was formed, contact holes were formed therein, and metal wirings 313 and 314 were formed. Thus, a pair of first regions having a high oxygen concentration can be formed outside the channel formation region, and a second region having a high oxygen concentration and constituting the source and drain can be formed outside the channel forming region. It was. Further, an N ⁻ -type region may be provided between the first region and the second region. (Fig. 5 (E))

  In this manner, a TFT having a structure similar to that shown in FIG. In this example, since the channel formation region was not doped with oxygen, a TFT having a large on / off ratio was obtained compared to the case of Example 1 and Example 2. In particular, even when the drain voltage was as large as + 10V, good characteristics without leakage current were obtained when the gate voltage was between 0 and -10V.

The LDD structure may be configured as follows. The gate electrode anodic oxide 309 is formed to a thickness of 50 to 100 nm. First, an N-type impurity is introduced in an amount of 1 × 10 ¹³ to 1 × 10 ¹⁴ cm ⁻² by means such as an ion implantation method. Furthermore, anodization may be performed again, the thickness may be 200 to 500 nm, and impurities of the same conductivity type may be added at a dose of 1 × 10 ¹⁵ to 1 × 10 ¹⁶ cm ⁻² . Thus, an LDD region or a substantially intrinsic region can be formed between the channel formation region and the source / drain regions. In particular, since oxygen also acts as an N-type impurity, phosphorus may be further added at 1 × 10 ¹³ to 1 × 10 ¹⁴ cm ⁻² .

The present embodiment will be described with reference to FIG. As a substrate, Corning 7059 was used, and a base silicon oxide film 402, an active layer 403, a silicon oxide gate insulating film 404, and an aluminum gate electrode 405 were formed on the substrate 401 by the same process as in Example 3. Then, by ion doping, carbon, nitrogen, oxygen, for example, oxygen ions were implanted in a self-aligning manner using the gate electrode 405 as a mask to form a region having a high oxygen concentration in the active layer. The dose of oxygen is 1 × 10 ^{14 to} 3 × 10 ¹⁶ cm ⁻² , for example, 5 × 10 ¹⁵ cm ⁻² so that the concentration is 1 × 10 ¹⁹ to 2 × 10 ²¹ cm ^−3. The energy was selected so that a peak would appear at the boundary between the silicon oxide film 404 and the gate insulating film. As a result, a region having oxygen with a concentration of 5 × 10 ^{19 to} 5 × 10 ²⁰ cm ⁻³ (depending on the depth) was formed in the active layer in the region where the source, drain, and junction were formed. (Fig. 6 (A))

Thereafter, the gate electrode is anodized in the same manner as in Example 1 to form an anodic oxide layer 406, and in the same manner as in Example 1, the gate electrode 405 and its anodic oxide layer 406 are used as a mask in the active layer in a self-aligned manner. Impurities (phosphorus) were doped to form impurity regions 407 and 408. The dose amount was smaller than that of Example 3 and was set to 1 × 10 ^{13 to} 5 × 10 ¹⁴ cm ⁻² , for example, 2 × 10 ¹⁴ cm ⁻² . And this was activated by laser irradiation. As a result, the impurity concentration of the source / drain was 1 × 10 ^{18 to} 5 × 10 ¹⁹ cm ⁻³ . (Fig. 6 (B))

  Subsequently, a silicon oxide film 409 having a thickness of 300 to 3000 nm, for example, 900 nm was formed by a plasma CVD method. (FIG. 6C) Next, this silicon oxide film 409 was etched by performing anisotropic dry etching by a known RIE method. At this time, on the side surface of the gate electrode 506 having a height of 700 nm, the thickness in the height direction is about twice the film thickness (the film thickness of the silicon oxide film is 900 nm). At this time, the silicon oxide film 404 that is a gate insulating film was also continuously etched to expose the source / drain regions 407 and 408.

  As a result of the above processes, insulators 410 and 411 having a substantially triangular shape remain on the side surface of the gate electrode. Thereafter, a film made of a material suitable for forming silicide such as titanium, tungsten, platinum, palladium, nickel, or the like having a thickness of 5 to 50 nm, for example, a titanium film 412 was formed by a sputtering method. (FIG. 6 (D)) Next, this is preliminarily heated to 250 to 450 ° C., irradiated with laser light, instantaneously heated to 500 to 800 ° C., and titanium and silicon are reacted to form titanium silicide. Silicide regions 413 and 414 are formed on the impurity regions (source / drain). Since titanium silicide has a low resistivity of 30 to 100 μΩ · cm, the sheet resistance of the substantial source and drain regions was 10 Ω / □ or less. At this time, if the diffusion of the silicide is performed until it reaches the glass substrate, a barrier layer becomes unnecessary in the contact portions of the aluminum electrodes 416 and 417 thereafter, and the process can be simplified.

This step may be performed by infrared lamp annealing (RTA). When lamp annealing is performed, lamp irradiation is performed for several minutes at 600 ° C. and for several seconds at 1000 ° C. (silicon wafer monitor temperature) so that the surface to be irradiated is about 600 to 1000 ° C. To do. Here _is the relationship _{E I <E O ≦ E N} ( relationship _{N I <N O ≦ N N} ).

  Thereafter, the Ti film that did not become silicide was removed by etching with an etching solution in which hydrogen peroxide, ammonia, and water were mixed at a ratio of 5: 2: 2. At this time, the titanium silicide layers 413 and 414 are not etched and can remain. A silicon oxide film having a thickness of 500 nm was formed as an interlayer insulator 415 by plasma CVD on the entire surface, contact holes were formed in the source / drain of the TFT, and aluminum wiring / electrodes 416 and 417 were formed. The TFT was completed through the above steps. In this example, since the aluminum electrodes 416 and 417 and the silicon of the source / drain are not in direct contact with each other, a very good contact was obtained. Thus, the TFT was completed. In order to activate the impurity region, hydrogen annealing may be further performed at 200 to 400 ° C. Thus, a pair of oxygen layers was formed outside the channel formation region and the silicide layer was formed as a pair in the source / drain region outside the channel formation region. (Fig. 5 (E))

  In the TFT shown in this embodiment, since the substantial source / drain sheet resistance is reduced by the silicide regions 413 and 414, the doping amount to the source / drain is 1 as compared with the case of the third embodiment. / 10 or less. For this reason, the electric field between the channel formation region and the drain region becomes gentle, and the TFT has a smaller off-current due to the effects of the present invention. Moreover, the time of the doping process could be reduced to 1/10 or less of the conventional time by reducing the dose.

  In the present invention, with respect to reducing the sheet resistance of the source / drain, in addition to the above, it is effective to increase the impurity concentration of the source / drain or to use silicide together. In addition, it goes without saying that the active layer is not limited to the solid phase growth method, and laser annealing, RTA, or the like may be used, or it may be used in combination with the solid phase growth method.

The structure and characteristics of the TFT of the present invention are shown. The structure and characteristics of a conventional TFT are shown. A manufacturing process of the TFT of Example 1 will be described. The manufacturing process of TFT of Example 2 is shown. The manufacturing process of TFT of Example 3 is shown. The manufacturing process of TFT of Example 4 is shown.

Explanation of symbols

1 Gate electrode 2 Gate insulating film 3 Impurity region (first region)
4 Second region 5 Channel formation region (third region)
6 Second region 7 Impurity region (first region)
11 Gate electrode 12 Gate insulating film 13 Impurity region 14 Offset region 15 Channel formation region (third region)
16 Offset region 17 Impurity region (first region)

Claims (10)

An active layer, a gate insulating film, a gate electrode overlapping with the active layer through the gate insulating film, an interlayer insulator on the active layer, and the active layer provided on the interlayer insulator on an insulating surface An electrode in contact with the layer,
The semiconductor device is characterized in that the active layer is selectively silicided from the surface in contact with the electrode. On the insulating surface, an active layer having a silicide region, a gate insulating film, a gate electrode overlapping with the active layer through the gate insulating film, an interlayer insulator on the active layer, and an interlayer insulator An electrode provided in contact with the active layer,
2. The semiconductor device according to claim 1, wherein the silicide region is formed by a reaction between the electrode and the active layer. In claim 2,
The semiconductor device, wherein the electrode is formed of a multilayer film of chromium and aluminum. On the insulating surface, an active layer, a gate insulating film, and a gate electrode that overlaps the active layer via the gate insulating film are formed,
Forming an interlayer insulator on the active layer;
Forming an electrode on the interlayer insulator so as to be in contact with the active layer;
A method for manufacturing a semiconductor device, wherein a silicide region is selectively formed below a surface of the active layer in contact with the electrode by reacting the electrode with the active layer. In claim 4,
The method for manufacturing a semiconductor device, wherein the electrode is formed of a multilayer film of chromium and aluminum. In claim 4 or claim 5,
A method for manufacturing a semiconductor device, wherein the electrode and the active layer are reacted by heating. On the insulating surface, an active layer, a gate insulating film, and a gate electrode that overlaps the active layer via the gate insulating film are formed,
Forming an interlayer insulator on the active layer;
Forming a film of titanium or nickel on the interlayer insulator so as to be in contact with the active layer;
By reacting the titanium or nickel film and the active layer, a silicide region is selectively formed below the surface of the active layer in contact with the titanium or nickel film,
After the silicide region is formed, the film made of titanium or nickel is removed, and an electrode is formed so as to be in contact with the silicide region. In claim 7,
The method for manufacturing a semiconductor device, wherein the electrode is formed of a multilayer film of titanium nitride and aluminum. In claim 7 or claim 8,
A method for manufacturing a semiconductor device, comprising: reacting a film made of titanium or nickel with the active layer by irradiating a laser. In claim 7 or claim 8,
A method for manufacturing a semiconductor device, comprising reacting the film made of titanium or nickel with the active layer by irradiating strong light of visible or near infrared light.

Images