JP3860148B2 - Manufacturing method of semiconductor circuit - Google Patents

Description

  The present invention has a matrix structure, such as a liquid crystal display device and a dynamic RAM (DRAM), and has a MOS type or MIS (metal-insulator-semiconductor) type field effect element (the above is referred to as a MOS type element) as a switching element. A matrix device (including an electro-optic display device and a semiconductor memory device), and a method for manufacturing a driver circuit therefor. In particular, the present invention relates to a method for manufacturing a device using a thin film semiconductor element such as a thin film semiconductor transistor formed on an insulating substrate as a MOS type element.

  Recently, research has been conducted on an insulating gate type semiconductor device having a thin-film active layer (also referred to as an active region) on an insulating substrate. In particular, thin-film insulated gate transistors, so-called thin film transistors (TFTs), have been eagerly studied. These are intended to be used for controlling each pixel in a display device such as a liquid crystal having a matrix structure, and are distinguished as an amorphous silicon TFT or a polycrystalline silicon TFT depending on the material and crystal state of the semiconductor to be used. Has been. However, recently, research has also been made to use a material that exhibits an intermediate state between polycrystalline silicon and amorphous. This material is said to be semi-amorphous, and is considered to be in a state where small crystals float in an amorphous structure. As will be described later, this material is an excellent material having the characteristics of high mobility in a single crystal state and low leakage current in an amorphous state.

  Also in a single crystal silicon integrated circuit, a polycrystalline silicon TFT is used as a so-called SOI technology, and this is used as a load transistor in a high integration SRAM, for example. However, in this case, the amorphous silicon TFT is hardly used.

  Furthermore, since a semiconductor circuit on an insulating substrate has no capacitive coupling between the substrate and the wiring, it can operate at a very high speed, and a technique for use as an ultra-high-speed microprocessor or an ultra-high-speed memory has been proposed.

  In general, the electric field mobility of an amorphous semiconductor is small, and therefore it cannot be used for a TFT that requires high-speed operation. In addition, since the P-type field mobility is extremely small in amorphous silicon, a P-channel TFT (PMOS TFT) cannot be manufactured. Therefore, in combination with an N-channel TFT (NMOS TFT), A complementary MOS circuit (CMOS) cannot be formed.

  However, a TFT formed of an amorphous semiconductor has a feature that the OFF current is small. Therefore, unlike a liquid crystal active matrix transistor, such a high-speed operation is not required, and only one of the conductivity types is sufficient, and it is used for an application that requires a TFT having a high charge retention capability. .

On the other hand, a polycrystalline semiconductor has a larger electric field mobility than an amorphous semiconductor, and thus can operate at high speed. For example, in a TFT using a silicon film recrystallized by laser annealing, a value of 300 cm ² / Vs is obtained as the electric field mobility. Since the electric field mobility of a MOS transistor formed on a normal single crystal silicon substrate is about 500 cm ² / Vs, the MOS circuit on the single crystal silicon has a very large value, and the parasitic capacitance between the substrate and the wiring However, since it is on an insulating substrate, there is no such limitation, and remarkable high-speed operation is expected.

  In addition, in the case of polycrystalline silicon, not only an NMOS TFT but also a PMOS TFT can be obtained in the same manner, so that a CMOS circuit can be formed. For example, in an active matrix liquid crystal display device, only an active matrix portion can be formed. In addition, those having a so-called monolithic structure in which peripheral circuits (drivers and the like) are also composed of CMOS polycrystalline TFTs are known.

  The TFT used in the above-described SRAM is also paying attention to this point, and the PMOS is constituted by a TFT, which is used as a load transistor.

  In addition, in a normal amorphous TFT, it is difficult to form a source / drain region by a self-alignment process such as that used in single crystal IC technology, which is due to a geometrical overlap between the gate electrode and the source / drain region. While the parasitic capacitance becomes a problem, the polycrystalline TFT has a feature that the parasitic capacitance can be remarkably suppressed because a self-alignment process can be adopted.

  Several problems have been pointed out with respect to the advantages of the polycrystalline TFT having such characteristics. A general polycrystalline TFT is a coplanar type in which an active layer is formed on an insulating substrate and a gate insulating film and a gate electrode are formed thereon. Although this structure has an advantage that a self-alignment process can be adopted, it is difficult to reduce the leakage current (OFF current) of the active layer.

  The details of the cause of this leakage current are not clear, but the major cause was due to the interfacial charge generated between the base and the active layer. Therefore, the problem was solved by paying close attention to the production of this interface and reducing the interface state density to the same level as that between the gate oxide film and the active layer.

  That is, in a high-temperature process (maximum process temperature of about 1000 ° C.), quartz is used as a substrate, a silicon film is formed thereon, and this is thermally oxidized at about 1000 ° C. to obtain a clean surface. After the formation, an active silicon layer was formed by a film forming method such as a low pressure CVD method.

  In a low-temperature process (a process having a maximum process temperature of 650 ° C. or lower, also referred to as an intermediate temperature process), a silicon oxide film having a low interface state density is formed as a base film between the substrate and the active layer as much as the gate insulating film. The method of doing was adopted. As a method for forming a silicon oxide film, a sputtering method is excellent. In addition, an oxide film having excellent characteristics can also be obtained by ECR-CVD or TEOS plasma CVD.

  However, the leakage current could not be improved. In particular, NMOS was more than an order of magnitude larger than PMOS. The inventor presumed that the cause was the N-type with a weak active layer. In fact, the phenomenon that the threshold voltages of PMOS and NMOS fabricated in a high temperature process or a low temperature process shift in the negative direction was observed with good reproducibility. It was speculated that this was due to the weak N-type when the crystallinity was not good like amorphous silicon, especially in the case of high purity with no other impurities added to silicon. It was speculated that high-temperature process polycrystalline silicon, unlike perfect single-crystal silicon, has many lattice defects and dangling bonds, which serve as donors and supply electrons. Of course, the possibility of the influence of a trace amount of mixed elements (such as sodium) remains.

Anyway, if there is such a cause, it can be explained that the threshold voltage of the NMOS is significantly lower than that of the PMOS and the leakage current is large. This is shown in FIG. In NMOS, and grounding the source 12 (N ⁺ -type) as shown in FIG. 1 (A), than the threshold voltage V _th to the gate electrode 11 in a state in which the drain 13 (N ⁺ -type) was applied a positive voltage When a large voltage is applied, a channel is formed on the gate electrode side of the active layer 14 and a drain current (solid arrow in the figure) flows. However, since the active layer 14 is a weak N type (N ⁻ type), the source A current (dotted arrow in the figure) that hardly depends on the gate voltage flows from the drain to the drain.

Even if the potential of the gate electrode is equal to or lower than the threshold voltage V _th , the dotted line current flows. When the potential of the gate electrode becomes a large negative value, an inversion layer (P-type) 16 is generated as shown in FIG. 1B. However, the entire channel does not invert, and an excessive voltage is applied. Then, electrons are accumulated on the opposite side of the gate and a channel is formed. The NMOS data actually obtained is consistent with this consideration.

On the other hand, in the PMOS, the threshold voltage increases because the active layer is N ⁻ type. However, the leak on the other side of the gate is greatly reduced. FIG. 2 shows a state in which a voltage lower than the threshold value or a voltage higher than the threshold value is applied to the PMOS.

  Such a remarkable leakage current in NMOS has become an obstacle in various application fields, particularly in fields requiring dynamic operation. For example, in a liquid crystal active matrix or a DRAM, image information and stored information are lost due to leakage current. Therefore, it has been necessary to reduce such a leakage current.

  One method is to make the NMOS active layer intrinsic (I-type) or weak P-type. For example, when an active layer of NMOS is made into an I-type or a weak P-type by implanting an appropriate amount of P-type impurity (for example, boron) into only NMOS or both NMOS and PMOS at the time of forming the active layer, The threshold voltage should rise and the leakage current should also be greatly reduced. However, this method has several problems.

  Normally, a CMOS circuit in which both NMOS and PMOS are mounted on a single substrate is used. However, if an impurity is implanted only in the N-type, an extra photolithography process is required. Further, if a P-type impurity is to be implanted into both the NMOS and PMOS active layers, a delicate impurity implantation technique is required. If the amount of implantation is too large, the PMOS threshold voltage will be decreased and the leakage current will increase.

  Ion implantation technology is also a problem. In the injection technique for performing mass separation, only necessary impurity elements can be injected, but the processing area is small. Further, in the so-called ion doping method, the processing area is large, but since there is no mass separation step, unnecessary ions are implanted, and the doping amount may not be accurate.

  Further, such a method of accelerating and implanting ions causes the formation of localized levels at the interface between the active layer and the base. In addition, unlike conventional ion implantation for single crystal semiconductors, implantation is performed on an insulating substrate, so that the charge-up phenomenon is serious and it is difficult to precisely control the implantation amount.

  Therefore, it is conceivable to mix P-type impurities in advance during the formation of the active layer, but it is difficult to control the amount of trace impurities, and when forming the NMOS and PMOS from the same film, If the amount is not appropriate, the leakage current of the PMOS is increased, and when the NMOS and the PMOS are formed from different films, an extra mask process is required. In addition, controlling the threshold voltage by such a method also causes variations in the threshold value of the TFT due to factors such as the gas flow rate, and the variation in the threshold value from lot to lot becomes remarkably large.

The present invention is intended to provide an answer to such a difficult problem. The main point of the present invention is to optimize the circuit design instead of reducing the leakage current of the NMOS depending on the process. Therefore, a circuit that can be used even with a TFT having a large leakage current is designed. As described above, when the active layer is formed from a high-purity silicon material, it becomes N ⁻ type, but its energy level is extremely reproducible and stable. In addition, the process itself is extremely simple and the yield is sufficiently high. On the other hand, various methods for controlling the threshold voltage not only complicate the process, but also the energy levels (Fermi level, etc.) of the obtained active layer vary from lot to lot, resulting in a decrease in yield. To do.

Obviously, the process of removing impurities as much as possible is easier than adjusting the NMOS to the circuit by improving the process, that is, performing a fine doping of about 10 ¹⁷ cm ⁻³ , and the resulting NMOS It is better to design the circuit according to your needs. This is the technical idea of the present invention.

In the present invention, a silicon nitride film and a silicon oxide film on the silicon nitride film are continuously formed on a glass substrate, and two or three P-channel types connected in series on the silicon oxide film. A semiconductor circuit manufacturing method for forming a thin film transistor, wherein the two or three P-channel thin film transistors each have a leakage current of 10 ⁻¹² A or less when the drain voltage is 1 V. This is a method for manufacturing a circuit.

In the present invention, a silicon nitride film and a silicon oxide film on the silicon nitride film are continuously formed on a glass substrate, a semiconductor film is formed on the silicon oxide film, and the semiconductor film is used. A method of manufacturing a semiconductor circuit for forming an active layer of two or three P-channel thin film transistors connected in series, wherein the two or three P-channel thin film transistors each have a drain voltage of 1V. The method for manufacturing a semiconductor circuit is characterized in that the leakage current is 10 ⁻¹² A or less.

In the present invention, a silicon nitride film and a silicon oxide film on the silicon nitride film are continuously formed on a glass substrate made of low alkali glass or non-alkali glass, and connected in series on the silicon oxide film. A method of manufacturing a semiconductor circuit in which two or three P-channel thin film transistors are formed, and each of the two or three P-channel thin film transistors has a leakage current of 10 ⁻¹² A when the drain voltage is 1V. A method for manufacturing a semiconductor circuit, characterized by the following.

In the present invention, a silicon nitride film and a silicon oxide film on the silicon nitride film are continuously formed on a glass substrate made of low alkali glass or non-alkali glass, and a semiconductor film is formed on the silicon oxide film. A method of manufacturing a semiconductor circuit in which an active layer of two or three P-channel thin film transistors connected in series is formed using the semiconductor film, wherein the two or three P-channel thin film transistors are In this method, a leakage current is 10 ⁻¹² A or less when the drain voltage is 1 V, respectively.

  The silicon nitride film may be formed using any one of a low pressure CVD method, a sputtering method, and a plasma CVD method, and the silicon oxide film may be formed using any one of a low pressure CVD method, a sputtering method, and a plasma CVD method. The silicon oxide film may be formed using TEOS.

  The silicon nitride film may have a thickness of 5 to 200 nm. The thickness of the silicon oxide film can be 20 nm to 1 μm.

  The two or three P-channel thin film transistors may be pixel driving transistors.

  The semiconductor circuit to which the present invention is applied is not universal. In particular, the present invention uses materials that change light transmittance and reflectivity due to the effect of an electric field, such as a liquid crystal display device, and these materials are sandwiched between opposing electrodes, and an electric field is applied between the opposing electrodes. In addition, an active matrix circuit for displaying an image, a memory device that retains memory by accumulating electric charge in a capacitor such as a DRAM, a MOS structure of a MOS transistor as a capacitor, or other capacitors It is suitable for a circuit having a dynamic circuit such as a dynamic shift register for driving a circuit of the next stage. In particular, the invention is suitable for a circuit in which a dynamic circuit and a static circuit are mixedly mounted.

  One example of the present invention is to use a PMOS TFT as a switching transistor in a display portion of an active matrix circuit such as a liquid crystal. Here, it is necessary that the PMOS TFT is inserted in series with the data line and the pixel electrode, and if the NMOS TFT is inserted in parallel, there is a large amount of leakage current. Is inappropriate. Therefore, the present invention includes a case where PMOS and NMOS TFTs are inserted in series in the pixel TFT circuit. Of course, it is also within the technical scope of the present invention that two PMOS TFTs are inserted in parallel.

  The second example of the present invention is that the driving circuit is a CMOS circuit in the device having the display circuit portion (active matrix) and the driving circuit (peripheral circuit) as described above. In this case, all the circuits need not be CMOS, but it is desirable that the transmission gate and the inverter circuit be CMOS. A conceptual diagram of such an apparatus is shown in FIG. In the figure, a data driver 31 and a gate driver 32 are formed on an insulating substrate 37, and an active matrix 33 having a PMOS TFT is formed in the central portion. These driver portion and active matrix are connected to a gate line 35 and data. A display device connected by line 36 is shown. The active matrix 33 is an aggregate of pixel cells 34 having PMOS.

  Regarding the CMOS circuit, for example, even if the threshold voltage of the obtained TFT is 2V for NMOS, 6V for PMOS, and the leakage current is more than 10 times higher for NMOS than for PMOS, there is no problem in the CMOS inverter. Absent.

  This is because in a logic circuit such as an inverter, power consumption due to leakage is not a problem. In addition, the operation of the inverter is required to be lower than the NMOS threshold voltage in the low voltage state and higher than the sum of the drain voltage and the PMOS threshold voltage (<0) in the high voltage state. There is no problem if the drain voltage is 8 V or higher, ideally 10 V or higher. For example, it is sufficient that the input has a binary value of 0 V and 8 V.

  The third example of the present invention relates to a semiconductor memory such as a DRAM. Semiconductor memory devices have already reached speed limits in single crystal ICs. In order to operate at higher speeds, it is necessary to increase the current capacity of the transistor, which not only causes a further increase in current consumption, but also stores memory by storing electric charge in the capacitor. As for the DRAM that performs the operation, there is no other way but to cope with it by increasing the drive voltage because the capacitance of the capacitor cannot be increased any more.

  The single crystal IC is said to have reached the speed limit because, in part, a large loss is caused by the capacitance of the substrate and the wiring. If an insulator is used for the substrate, it can be driven at a sufficiently high speed without increasing current consumption. For these reasons, an IC having an SOI (semiconductor on insulator) structure has been proposed.

  Even in the case of a DRAM, in the case of a 1Tr / cell structure, the circuit configuration is almost the same as that of the previous liquid crystal display device. A PMOS TFT with a small current is used. The basic block configuration is the same as that of FIG. For example, in a DRAM, 31 is a column decoder, 32 is a row decoder, 33 is a storage element section, 34 is a unit storage bit, 35 is a bit line, 36 is a word line, and 37 is an (insulating) substrate.

  Both the active matrix and the DRAM of the liquid crystal display device require a refresh operation, but during the refresh period, the charge accumulated in the pixel capacity and capacitor capacity is not discharged. Thus, the TFT needs to function as a sufficiently large resistance. If an NMOS TFT is used in this case, sufficient driving cannot be performed because of a large leakage current. This is the advantage of using a PMOS TFT with low leakage current.

In the present invention, a high temperature process TFT is effective, but a low temperature process TFT is particularly effective. A TFT obtained by a low-temperature process has a structure of an active layer that is intermediate between an amorphous and a single crystal, has a large lattice distortion, and is in a so-called semi-amorphous state, and thus is physically close to an amorphous state. That is, the active layer produced by a pure silicon material by a low temperature process is usually N ^- type.

  Here, when the semi-amorphous state is described in detail, the silicon in the amorphous state starts crystal growth as heat is applied, but it is not in a state of crystal growth up to about 650 ° C. under atmospheric pressure. That is, there is a relatively poor crystallinity portion between good crystallinity portions, and the intermolecular bond is tight, which is different from the crystal precipitation in ordinary ionic crystals. In other words, the number of dangling bonds is extremely small. If the crystal growth exceeds 680 ° C., the crystal growth rate is remarkably accelerated and a polycrystalline state composed of many crystal grains is obtained. In this case, the molecular bond at the crystal grain boundary that has been buffered by the lattice strain is broken, and a large number of dangling bonds are formed at the grain boundary part.

  In such a semi-amorphous material, even if an impurity is doped in the active layer, it does not contribute much to activation as in the case of amorphous silicon. As a cause of this, the present inventors believe that the dopant impurities are selectively trapped in a portion where there are particularly many dangling bonds. Therefore, in a semi-amorphous active layer or an active layer formed by a low temperature process, it is difficult to control the threshold voltage by doping.

The present invention is also effective in a TFT having two active layers as described in Japanese Patent Application No. 4-73315 which is the invention of the present inventors. In this TFT, an active layer in an amorphous state is provided on the substrate side, and an active layer in a semi-amorphous or polycrystalline state is provided on the TFT. Leakage caused by charges existing at the interface between the substrate and the active layer is minimized. Can be reduced. However, because the structure uses amorphous silicon, the lower active layer is N ⁻ type. Therefore, even if the leakage due to the interface can be reduced, the leakage due to the active layer cannot be easily reduced. For example, even if the leakage current is 10 ^-12 A or less (drain voltage 1 V) in PMOS, the leakage current is 100 times or more in NMOS.

  The manufacturing method is illustrated in FIG. First, a film 42 having a strong passivation force such as silicon nitride is formed on the substrate 41. If the substrate is sufficiently clean, such a film need not be formed. Further, a base oxide film 43 is formed. Then, two layers of amorphous silicon film are formed. By optimizing the deposition rate and the deposition substrate temperature, it is determined whether the amorphous state will be maintained, semi-amorphized or polycrystallized by the subsequent heat treatment. The In this example, the upper layers 45 and 47 are semi-amorphized (or polycrystallized), and the lower layers 44 and 47 remain amorphous.

A feature of such a method is that silicon films having two different properties can be formed by slightly changing the conditions while performing film formation using the same chamber. Voltage control will undermine the advantages of this method. Even if the lower layers 44 and 46 are changed from N ^- type to I-type, this layer remains amorphous, so the ionization rate is poor and a large amount of doping is required. Therefore, there is a possibility that the chamber is significantly contaminated by these impurities, and conversely, the PMOS active layer is made P-type. Therefore, a TFT having such a two-layer active layer is extremely suitable for the present invention which does not require threshold voltage control by doping. A method for forming such a TFT will be described in detail in Examples.

  The invention has made it possible to increase the reliability and performance of particularly dynamic circuits and devices having such circuits. Conventionally, polycrystalline TFTs have a low ON / OFF ratio and have various difficulties in practical use, especially for purposes such as active matrix of liquid crystal display devices, but such problems are almost solved by the present invention. Seems to have been. Furthermore, as shown in Embodiment 2, the semiconductor circuit on the insulating substrate is excellent in terms of high-speed operation. Although not shown in the embodiments, it will be apparent that the present invention can be effectively applied to a TFT used as a means for three-dimensionalization of a single crystal semiconductor integrated circuit.

  For example, the peripheral logic circuit may be formed of a semiconductor circuit over a single crystal semiconductor, and a TFT may be provided thereon via an interlayer insulator, thereby forming a memory element portion. In this case, the memory element portion is a DRAM circuit using PMOS TFTs, and the drive circuit is configured as a single crystal semiconductor circuit in CMOS. In addition, when such a circuit is used for a microprocessor, the memory section is raised to the second floor, so that the area can be saved. Thus, the present invention is considered to be an extremely useful invention in industry.

  FIG. 4 illustrates an example of manufacturing a CMOS circuit using the present invention. In this example, a Corning 7059 glass substrate was used as the substrate 41. In addition to these, various types of substrates can be used, but it is necessary to deal with the substrate so that mobile ions such as sodium do not enter the semiconductor coating. An ideal substrate is a synthetic quartz substrate having a low alkali concentration. However, when it is difficult to use it in terms of cost, a commercially available low alkali glass or non-alkali crow is used. In this embodiment, a silicon nitride film 42 having a thickness of 5 to 200 nm, for example, 10 nm, is formed on the substrate 41 by a low pressure CVD method in order to prevent intrusion of mobile ions from the substrate. Further, a silicon oxide film 43 having a thickness of 20 to 1000 nm, for example, 50 nm was formed on the silicon nitride film by sputtering. The thickness of these coatings is designed according to the degree of penetration of mobile ions or the degree of influence on the active layer.

  For example, when the quality of the silicon nitride film 42 is not good and the charge trap is large, the upper semiconductor layer is affected through the silicon oxide film. In this case, the silicon oxide film 43 needs to be thickened.

  These films may be formed not only by the low pressure CVD method and the sputtering method as described above, but also by a method such as a plasma CVD method. In particular, TEOS may be used for forming the silicon oxide film. The selection of these means may be determined in consideration of the investment scale and mass productivity. Needless to say, these films may be formed continuously.

  Thereafter, an amorphous silicon film having a thickness of 20 to 200 nm, for example, 100 nm, was formed by mono-silane as a raw material by a low pressure CVD method. The substrate temperature was set to 430 to 480 ° C., for example, 450 ° C. Furthermore, the substrate temperature was continuously changed to form an amorphous silicon film having a thickness of 5 to 200 nm, for example 10 nm, at 520 to 560 ° C., for example, 550 ° C. As a result of the inventors' study, it has been clarified that the substrate temperature has an important influence on the subsequent crystallization. For example, a film formed at 480 ° C. or lower was difficult to crystallize. Conversely, the film formed at a temperature of 520 ° C. or higher was easy to crystallize. The amorphous silicon film thus obtained was thermally annealed at 600 ° C. for 24 hours. As a result, only the upper silicon film was crystallized to obtain crystalline silicon called so-called semi-amorphous silicon. On the other hand, the lower silicon film remained in an amorphous state.

In order to promote crystallization of the upper silicon film, it is desirable that the concentrations of carbon, nitrogen, and oxygen contained in the film are all 7 × 10 ¹⁹ cm ⁻³ or less. In this example, it was confirmed by SIMS analysis that it was 1 × 10 ¹⁷ cm ⁻³ or less. Conversely, in order to suppress crystallization of the lower silicon film, it is convenient that many of these elements are contained. However, since excessive doping adversely affects the semiconductor characteristics, and thus the TFT characteristics, the presence / absence of doping and the amount thereof are designed according to the characteristics of the TFT.

Now, after the amorphous silicon film is made into a crystalline silicon film by thermal annealing, it is etched into an appropriate pattern to form an island-shaped semiconductor region 45 for NTFT and an island-shaped semiconductor region 47 for PTFT. It was confirmed by SIMS (secondary ion mass spectrometry) that the upper part of each island-like semiconductor region was not intentionally doped with impurities, and that the impurity concentration of boron or the like was 10 ¹⁷ cm ⁻³ or less. Therefore, the conductivity type of this part is presumed to be N ⁻ type. On the other hand, the silicon layers 44 and 46 below the respective semiconductor regions were substantially amorphous silicon.

  Thereafter, a gate insulating film (silicon oxide) 48 having a thickness of 50 to 300 nm, for example, 100 nm was formed by sputtering using silicon oxide as a target in an oxygen atmosphere. This thickness is determined by the operating conditions of the TFT.

  Next, an aluminum film having a thickness of 500 nm was formed by sputtering, and this was patterned with a mixed acid (a phosphoric acid solution added with 5% nitric acid) to form gate electrodes / wirings 49 and 50. The etching rate was 225 nm / min when the etching temperature was 40 ° C. In this way, the outer shape of the TFT was adjusted. The size of the channel at this time was 8 μm in length and 20 μm in width. The state at this time is shown in FIG.

  Furthermore, aluminum oxide was formed on the surface of the aluminum wiring by an anodic oxidation method. As an anodic oxidation method, the method described in Japanese Patent Application No. 3-231188 or Japanese Patent Application No. 3-238713 which is the invention of the present inventors was used. The detailed implementation may be changed depending on the characteristics of the target element, process conditions, investment scale, and the like. In this example, aluminum oxide films 51 and 52 having a thickness of 250 nm were formed by anodic oxidation.

Thereafter, an N-type source / drain region 53 and a P-type source / drain region 54 were formed by an ion implantation method through a gate oxide film, using a known CMOS fabrication technique. In either case, the impurity concentration was set to 8 × 10 ¹⁹ cm ⁻³ . As the ion source, boron fluoride ions were used for the P type and phosphorus ions were used for the N type. The former was implanted at an acceleration voltage of 80 keV and the latter at an acceleration voltage of 110 keV. The acceleration voltage is set in consideration of the thickness of the gate oxide film and the thickness of the semiconductor regions 45 and 47. An ion doping method may be used instead of the ion implantation method. In the ion implantation method, ions to be implanted are separated by mass, so unnecessary ions are not implanted, but the size of the substrate that can be processed by the ion implantation apparatus is limited. On the other hand, the ion doping method has the ability to process relatively large substrates (for example, 30 inches diagonal or more), but hydrogen ions and other unnecessary ions are simultaneously accelerated and implanted, so that the substrate is easily heated. . In this case, selective impurity implantation using a photoresist as a mask used in the ion implantation method is difficult.

  In this way, a TFT having an offset region was produced. This is shown in FIG. Finally, the source / drain regions were recrystallized by laser annealing using the gate electrode portion as a mask. As the conditions for laser annealing, for example, the methods described in Japanese Patent Application Nos. Hei 3-231188 and Hei 3-238713 were used. Then, silicon oxide was formed as an interlayer insulator 55 by the RF plasma CVD method, holes for electrode formation were made in this, and aluminum wirings 56 to 48 were formed to complete the device.

  In this embodiment, not only the coatings 45 and 47 that were originally crystalline silicon but also the coatings 44 and 46 that were amorphous silicon were crystallized by laser annealing. This is because laser annealing is powerful. As a result, as shown in FIG. 4C, the initial amorphous regions 44 and 46 are all converted to a material having the same crystallinity as the source / drain except for the lower portions 59 and 60 under the channel. As a result, the thickness of the source / drain was substantially the same as that of the island-shaped semiconductor regions 45 and 47. However, the substantial channel thickness was thinner than the source / drain regions, such as about 10 nm, as is apparent from the figure. As a result, the sheet resistance of the source / drain was small, and it was possible to show excellent characteristics that the OFF current was small because the channel was thin.

  FIG. 4 shows a manufacturing process of a CMOS circuit used for a driving circuit of a liquid crystal display device, and PMOSs are formed in the same manner in active matrix portions on the same substrate. The characteristics of the TFT thus formed were as follows: channel length was 5 μm, channel width was 20 μm, source / drain voltage was 1 V, NMOS leakage current was ˜100 pA, and PMOS was ˜1 pA of PMOS. . Thus, the off resistance was 100 times larger in the PMOS. When the gate voltage was + 8V (−8V in the case of PMOS), a current of 10 μA was applied to NMOS and 100 nA was supplied to PMOS. The reason why the drain current of PMOS is significantly smaller than that of NMOS is that when the threshold voltage is PMOS, it is shifted negatively. Therefore, when the gate voltage of the PMOS is set to -12V, the drain current is 1 μA. That is, if such a TFT is used to form a transmission gate, the potential applied to the PTFT should be shifted to the negative side.

The size of the PMOS TFT in the active matrix portion was set to a channel length of 5 μm and a channel width of 10 μm. When the gate voltage of the PMOS TFT used as the active matrix is changed from 0V to -12V, the drain current increases to 10 ⁶ times, so there is no problem for image display. Further, when it is necessary to make a large variation, it is preferable to construct two PMOS TFTs in series to form a so-called dual gate structure. In this case, in the OFF state, the resistance of the TFT further increases by about one digit. However, in the ON state, the resistance of the TFT only doubles, so that the drain current eventually varies by 10 ^7. . If TFTs are formed in three stages in series, the variation rate further increases by an order of magnitude.

FIG. 5 shows a manufacturing process of NMOS and PMOS devices for carrying out the present invention. In this example, a TFT manufactured by a high temperature process was manufactured. First, an undoped polysilicon film having a thickness of 100 to 500 nm, preferably 150 to 200 nm, was formed on a quartz substrate 61 (width 105 mm × length 105 mm × thickness 1.1 mm) by low pressure CVD. . This was oxidized in a dry high-temperature oxygen atmosphere. The temperature was in the range of 850 to 1100 ° C, with 950 to 1050 ° C being particularly preferred. In this manner, a silicon oxide film 62 was formed over the substrate (FIG. 5A).

  Further, an amorphous silicon film having a thickness of 100 to 1000 nm, preferably 350 to 700 nm, was formed by plasma CVD or low pressure CVD using disilane as a raw material. The substrate temperature was 350 to 450 ° C. Then, this was annealed for a long time at 550 to 650 ° C., preferably 580 to 620 ° C., to give crystallinity. Then, this was patterned to form an NMOS region 63a and a PMOS region 63b as shown in FIG.

  Subsequently, the surface of the silicon region 63 is oxidized in a dry high-temperature oxidizing atmosphere, and as shown in FIG. 5C, silicon oxide having a thickness of 50 to 150 nm, preferably 50 to 70 nm, is formed on the surface of the silicon region. A film 64 was formed. The oxidation conditions were the same as those for silicon oxide 62.

Thereafter, a silicon film doped with phosphorus of 10 ^{19 to} 2 × 10 ²⁰ cm ⁻³ , for example, 8 × 10 ¹⁹ cm ⁻³ is formed to a thickness of 200 to 500 nm, preferably 350 to 400 nm. ) To form an NMOS gate 65a and a PMOS gate 65b. Further, NMOS and PMOS impurity regions 66 and 67 were formed by ion implantation, respectively.

  At this time, the bottom surfaces of these impurities were prevented from reaching the underlying silicon oxide film 62. That is, many localized levels are formed at the interface between the underlying oxide film and the silicon film, and as a result, the silicon film near the underlying oxide film exhibits a specific conductivity type (usually N-type). If the impurity region is adjacent to such a portion of the silicon film, leakage occurs. Therefore, in order to avoid such a leak, a space of 50 to 200 nm is provided between the bottom surface of the impurity region and the base oxide film 62 in this embodiment.

  In this embodiment, ion implantation is performed through the silicon oxide film 64. However, in order to control the depth of the impurity region more precisely, the silicon oxide film 64 may be removed and thermal diffusion may be performed.

  After forming the impurity region, the crystallinity of the impurity region was recovered by thermal annealing. Thereafter, in the same manner as in a normal TFT manufacturing process, an interlayer insulator (phosphorus boron glass) 68 was deposited and planarized by reflow, and contact holes were formed to form metal wirings 69 to 71.

A 1Tr / cell DRAM (16 kbit) was fabricated using the TFT formed by the above process. When the TFT channel size is 2 μm channel length and 10 μm channel width, the NMOS leakage current is about 10 pA when the source / drain voltage is 1 V, and the PMOS leakage current is about 0 under the same conditions. 1 pA. As the memory element portion, a PMOS having a channel length of 2 μm and a channel width of 2 μm was used. The capacity of the capacitor in the memory element portion was 0.5 pF, and a long-term memory retention of a maximum refresh period of 5 seconds became possible. This was made possible because the PMOS off-state resistance was as high as 5 × 10 ¹³ Ω. Also, the peripheral circuit was made into CMOS using NMOS and PMOS manufactured in the above process. Since the DRAM is on such an insulating substrate, it can operate at high speed, and can be written and read at 100 nsec per bit.

The conceptual diagram of operation | movement of NMOS TFT is shown. A conceptual diagram of the operation of a PMOS TFT is shown. The conceptual diagram of the structure of this invention is shown. The manufacturing process of the TFT of the present invention is shown. The manufacturing process of the TFT of the present invention is shown.

Explanation of symbols

11, 21... Gate electrodes 12, 22... Source regions 13, 23... Drain regions 14, 24... Active layers 15, 25. Driver (column decoder in the case of DRAM)
32 ... Gate driver (low decoder in the case of DRAM)
33 ... Active matrix part (memory element part in the case of DRAM)
34: Unit pixel (unit storage bit in case of DRAM)
35 ... Gate line (bit line in case of DRAM)
36: Data line (in the case of DRAM, word line)
37 ... Insulating substrate

Claims (9)

A silicon nitride film and a silicon oxide film on the silicon nitride film are continuously formed on a glass substrate,
A method of manufacturing a semiconductor circuit in which two or three P-channel thin film transistors connected in series are formed on the silicon oxide film,
A thin film transistor located at an end of the two or three P-channel thin film transistors is connected to a pixel electrode,
In each of the two or three P-channel thin film transistors, the gate insulating film is on the channel forming region, the gate electrode is on the gate insulating film, and the channel forming region is an upper layer made of a crystalline semiconductor and an amorphous semiconductor. The source region and the drain region are crystalline semiconductors, and the upper layer of the channel formation region is thinner than the source region and the drain region. A method for manufacturing a semiconductor circuit. A silicon nitride film and a silicon oxide film on the silicon nitride film are continuously formed on a glass substrate,
Forming a semiconductor film on the silicon oxide film;
A method for manufacturing a semiconductor circuit, wherein an active layer of two or three P-channel thin film transistors connected in series is formed using the semiconductor film,
A thin film transistor located at an end of the two or three P-channel thin film transistors is connected to a pixel electrode,
In each of the two or three P-channel thin film transistors, the gate insulating film is on the channel forming region, the gate electrode is on the gate insulating film, and the channel forming region is an upper layer made of a crystalline semiconductor and an amorphous semiconductor. A portion of the channel formation region that is in contact with the silicon oxide film is an amorphous semiconductor, a source region and a drain region are crystalline semiconductors, and the upper layer of the channel formation region is A method for manufacturing a semiconductor circuit, wherein the thickness is thinner than the source region and the drain region . On a glass substrate made of low alkali glass or non-alkali glass, a silicon nitride film and a silicon oxide film on the silicon nitride film are continuously formed,
A method of manufacturing a semiconductor circuit in which two or three P-channel thin film transistors connected in series are formed on the silicon oxide film,
A thin film transistor located at an end of the two or three P-channel thin film transistors is connected to a pixel electrode,
In each of the two or three P-channel thin film transistors, the gate insulating film is on the channel forming region, the gate electrode is on the gate insulating film, and the channel forming region is an upper layer made of a crystalline semiconductor and an amorphous semiconductor. A portion of the channel formation region that is in contact with the silicon oxide film is an amorphous semiconductor, a source region and a drain region are crystalline semiconductors, and the upper layer of the channel formation region is A method for manufacturing a semiconductor circuit, wherein the thickness is thinner than the source region and the drain region . On a glass substrate made of low alkali glass or non-alkali glass, a silicon nitride film and a silicon oxide film on the silicon nitride film are continuously formed,
Forming a semiconductor film on the silicon oxide film;
A method for manufacturing a semiconductor circuit, wherein an active layer of two or three P-channel thin film transistors connected in series is formed using the semiconductor film,
A thin film transistor located at an end of the two or three P-channel thin film transistors is connected to a pixel electrode,
In each of the two or three P-channel thin film transistors, the gate insulating film is on the channel forming region, the gate electrode is on the gate insulating film, and the channel forming region is an upper layer made of a crystalline semiconductor and an amorphous semiconductor. The portion of the channel formation region that is in contact with the silicon oxide film is an amorphous semiconductor, the source region and the drain region are crystalline semiconductors, and the upper layer of the channel formation region is the upper layer of the channel formation region. A method for manufacturing a semiconductor circuit, wherein the thickness is thinner than the source region and the drain region .   5. The method for manufacturing a semiconductor circuit according to claim 1, wherein the silicon nitride film is formed using any one of a low pressure CVD method, a sputtering method, and a plasma CVD method.   6. The method for manufacturing a semiconductor circuit according to claim 1, wherein the silicon oxide film is formed using any one of a low pressure CVD method, a sputtering method, and a plasma CVD method.   6. The method for manufacturing a semiconductor circuit according to claim 1, wherein the silicon oxide film is formed using TEOS.   8. The method for manufacturing a semiconductor circuit according to claim 1, wherein the silicon nitride film has a thickness of 5 to 200 nm.   The method for manufacturing a semiconductor circuit according to claim 1, wherein the silicon oxide film has a thickness of 20 nm to 1 μm.

Images