Thin film transistor

[00002] This application is a Continuation of application Ser. No. 07/929,388, filed Aug. 14, 1992, now abandoned.

[00003] 1. Field of the Invention

[00004] The present invention relates to a thin film semiconductor device and a method for forming the same. More particularly, the present invention relates to a thin film semiconductor device and a method for forming the same which excel in confidence, productivity, and yield thereof. The present invention is applicable to e.g. a three dimensional integrated circuit or a driving circuit used in a liquid crystal display or a thin film image sensor and the like.

[00005] 2. Description of the Related Art

[00006] Heretofore, a semiconductor integrated circuit has been mainly a monolithic type, which has been formed on such semiconductor substrate as silicon. Recently, it has been tried to form it on an insulating substrate made of glass, sapphire, and the like. This is caused by the reasons that an operation speed will be raised according to the lowering of parasitic capacity between a substrate and a wiring, especially quartz or glass materials are not limited in size like a silicon wafer, cheap, easy to separate each device, and further a latch-up phenomenon, which poses a problem in a monolithic circuit of CMOS, will not arise. Also, it is necessary to form a thin film transistor (hereinafter refer to as TFT) etc. on a transparent substrate, because a semiconductor device and a liquid crystal element, or a light detecting element are required to be composed in a body, in a liquid crystal display and a contact type image sensor.

[00007] According to these reasons, a thin film semiconductor device has come to be formed on an insulating substrate. As shown in FIG. 5 which represents an instance of the usual thin film semiconductor device, a film 503 of silicon oxide etc. is formed on an insulating substrate 501, as a passivation film, and thereon, TFT is formed independent of other TFT. TFT has a source (drain) region 507, a drain (source) region 509, a channel forming region 508 (simply called as a channel region) which is held between the said regions of 507 and 509, a gate insulating film 504, a gate electrode 510, a source (drain) electrode 511, and a drain (source) electrode 512, just like MOSFET of a monolithic integrated circuit. Also, there is provided with an interlayer insulation 506 of PSG (phosphosilicate glass) so that a multilayer interconnect can be made.

[00008] The instance of FIG. 5 is called as a coplanar type, but, TFT has further different types such as an inverse coplanar type, a stagger type, and an inverse stagger type, according to the arrangement of a gate electrode and a channel region. These types are well known in the usual literatures to which the details can be referred.

[00009] Also in case of the monolithic integrated circuit, such alkaline ion as sodium and potassium, or such transition metal ion as iron, copper, and nickel poses a serious contamination problem, and much attention has been paid to stop the intrusion of these ions. In case of TFT, the problem of these ions are important alike, and close attention has been so given to the purification in a manufacturing process as to avoid the ion contamination to the utmost. And also a countermeasure has been taken so that the contamination of these ions do not extend to the device.

[00010] A thin film semiconductor device is different from a monolithic integrated circuit, in a point that a contaminated ion concentration in a substrate of the former is rather higher than that of the latter. That is, a single silicon crystal used in a monolithic integrated circuit has been so manufactured as to eliminate these harmful elements, based on the long experienced technologies. The content of these elements contaminated in the present products on sale is 10¹⁰ cm⁻³ or less.

[00011] However, in general, the concentration of elements contaminated in an insulating substrate usable for a thin film semiconductor device is not low. Of course, it is theoretically possible to reduce a strange element concentration, which will be the above contamination source, in case of such single crystal substrate as a spinel or a sapphire substrate. But it is not actual from a profitable point of view.

[00012] Also, in case of a quartz substrate, it is ideally possible to suppress the intrusion of the strange element, if it is prepared in a gaseous reaction (vapor reaction), using a high purity of silane gas and oxygen as a raw material. But it is difficult to send out the strange element to the outside in case it is once taken in, as the structure of quartz substrate is amorphous. A substrate used in a liquid crystal display gives a priority to the cost problem in particular, and then it is needed to use a cheap one, which originally contains various sorts of strange elements so as to make the manufacturing and the processing easy. There are strange elements which are not desirable in themselves for a semiconductor device, and there are cases where the strange elements will be mixed from the outside in the process of their adding, or these are contained in an adding material as an impurity.

[00013] Since for example, AN glass manufactured by Asahi Glass Corporation having as a main component barium borosilicate glass is a cheap glass substrate, heat resisting, and equivalent to silicon in a heat expansion coefficient, it is desirable for a substrate used in a liquid crystal display. But it contains about 5% of lithium, a portion of which is ionized to intrude into a semiconductor device as a mobile ion, and to deteriorate a device grade. Also it is difficult to prepare a high purity lithium of 99% or more, and about 0.7% of sodium is usually contained therein. An ionization ratio of sodium is very high to show about 10%, which brings about a very serious effect on the device features.

[00014] As shown in FIG. 5, in the conventional thin film semiconductor device, silicon oxide etc. has been used as a passivation film to cope with the intrusion of these mobile ions. Also these ions have been dealt with by gettering them, employing PSG (Phosphosilicate glass) or BPSG (Borophosphosilicate glass) as a interlayer insulation. However, it has been difficult to prevent the contamination effectively by these methods.

[00015] Accordingly, it is an object of the present invention to effectively suppress the deterioration of a device which is caused by the intrusion of these contaminated elements or ions. In accordance with the present invention, it is characterized in that so as to control the above contamination, a blocking film such as a film of silicon nitride, aluminum nitride, aluminum oxide, tantalum oxide and the like, which has a function to block the mobile ion, is formed on and beneath a thin film semiconductor device, and further, 1×10¹⁸ to 5×10²⁰ atoms/cm³, preferably 1×10¹⁹ to 1×10²⁰ atoms/cm³ of such halogen element as chlorine and fluorine is contained in either one or two of the semiconductor film (channel region provided in the semiconductor film) and the gate insulating film which constitute TFT. The halogen element has a function that it will combine with such mobile ion as sodium strongly, in the semiconductor film or the insulating film, and then will lower the ion effect exceedingly.

[00016] A typical instance according to the present invention is shown in FIG. 1, in which TFT applied by the present invention is illustrated. That is, the first silicon nitride film 102 is formed on an insulating substrate 101, as the first blocking film which functions to keep away the contamination through a substrate. On the first silicon nitride film, e.g. such silicon oxide film (insulating film) 103 as has a good contacting property with a silicon material is formed. If TFT is prepared by a direct forming of a semiconductor film, on the first silicon nitride, without forming the film 103, TFT will not operate, as a channel region will act as a conductor, according to a trap level generated in a boundary surface between the silicon nitride and the semiconductor material. It is therefore important to provide such a buffer material.

[00017] TFT is formed on the film 103. TFT has a source (drain) region 107, a drain (source) region 109, a channel region 108 provided between the regions of 107 and 109, a gate insulating film 104, and a gate electrode 110. Each region of source, drain, and channel of TFT is formed, employing a semiconductor material which is a single crystal, a polycrystal, or an amorphous. As the semiconductor material, for example, silicon, germanium, silicon carbide, and alloys thereof can be used.

[00018] Then, as the second blocking film, the second silicon nitride film 105 is formed covering this TFT. The present invention has the features that the second silicon nitride film is formed, after the formation of TFT and also before the formation of an electrode at the source and/or the drain. In the prior arts, a silicon nitride film as a final passivation film is formed, after a formation of an electrode. The silicon nitride film of the present invention, however, is different from that formed by the prior arts under such meaning. That is, the second silicon nitride film of the present invention is formed, in order to envelop TFT along with the first silicon nitride film. The present invention also aims at the prevention of contamination in an electrode formation process, after the forming of TFT. Therefore, it may be possible to form the silicon nitride film as a final passivation film, like the prior arts, after TFT and its accompanied electrode or wirings are formed in accordance with the present invention.

[00019] Then, after the formation of the second silicon nitride film, an interlayer insulating film 106 is formed, with an interlayer insulating material e.g. PSG (phosphosilicate glass) and so on. After that, a source (drain) electrode 111 and a drain (source) electrode 112 are formed. As aforementioned, it may be possible that an aluminum oxide, an aluminum nitride or a tantalum oxide instead of a silicon nitride is used as a blocking film.

[00020] In the instance of FIG. 1, however, a gate insulating film extends in a distance, and there is a possibility that a mobile ion etc. will intrude into the innerside of TFT from an edge part of the said gate insulating film extension. The improvement on this point is shown in FIG. 2, in which a gate insulating film is situated only on TFT. Then, this type has no problem posed in FIG. 1, but it is possible to disturb an operation of TFT in this case, on account of the reason that a silicon nitride part, which is in contact with a source region and a drain region parts being adjacent to a channel region, is polarized or traps electron by a gate voltage.

[00021] An instance which overcomes the above problem is shown in FIG. 3, in which a source region and a drain region being adjacent to a channel region are not neighboring to and not in contact with a silicon nitride film. So that the above problem of the polarization and the electron-trap of a silicon nitride can be solved. However, in case of applying a self-alignment process using a gate electrode as a mask, in the forming of a source and a drain region, it is necessary to implant an acceptor or a donor element through a gate insulating film, similarly in an instance of FIG. 1. For that purpose, if an ion implantation method is employed, it is required to raise an ion acceleration energy. On that occasion, a high speed ion will be implanted thereinto, and then, it is possible that a source and a drain region are expanded by a secondary scattering of the high speed ion.

[00022] In FIG. 2, the numerals are as follows:

[00023] 201; an insulating substrate, 202; the first silicon nitride film, 203; an insulating film used in buffering for silicon nitride etc. 204; a gate insulating film, 205; the second silicon nitride film, 206; an interlayer insulating film, 207; a source (drain) region, 208; a channel region, 209; a drain (source) region, 210; a gate electrode, 211; a source (drain) electrode, and 212; a drain (source) electrode.

[00024] In FIG. 3, the numerals are as follows:

[00025] 301; an insulating substrate, 302; the first silicon nitride film, 303; an insulating film used in buffering for silicon nitride etc., 304; a gate insulating film, 305; the second silicon nitride film, 306; an interlayer insulating film, 307; a source (drain) region, 308; a channel region, 309; a drain (source) region, 310; a gate electrode, 311; a source (drain) electrode, and 312; a drain (source) electrode.

[00026] In case where a silicon nitride film is used as a blocking film, according to the present invention, a good result has been obtained by selecting the followings.

[00027] In the formula represented by SiN_x, x=1.0 to 1.7, preferably x=1.3 to 1.35. (stoichiometric composition, x=1.33). Therefore, in the present invention, it is better to form a silicon nitride by using a low pressure CVD method. But it is needless to say that even the silicon nitride film formed by a plasma CVD method or a light CVD method has more confidence in a device quality, compared with the one formed without using the present invention.

[00028] In case where a silicon nitride film is formed by a low pressure CVD method, as a raw material, dichlorosilane (SiCl₂H₂) and ammonia (NH₃) are used. As a reaction condition, pressure: 10 1000 Pa and temperature: 500 to 800° C., preferably 550 to 750° C., are employed. Of course, silane (SiH₄) and tetrachlorosilane (SiCl₄) can be used.

[00029] In case where an aluminum oxide film or a tantalum oxide film is used in the present invention, the closer to Al₂O₃ or Ta₂O₅ a stoichiometric composition was, the better result could be obtained. These films are formed by a CVD method or a sputtering method. For example, an aluminum oxide film can be formed by oxidation of trimethyl aluminum Al(CH₃)₃ with nitrogen oxides (N₂O, NO, NO₂).

[00030] In order to carry out the present invention more effectively, it is recommended that the hydrogen atom concentration in a semiconductor film of such thin film semiconductor device as TFT be four times or less, preferably one time or less of the added halogen atom, and that the concentration of a harmful element such as carbon, nitrogen, or oxygen be 7×10¹⁹ cm⁻³ or less, preferably 1×10¹⁹ cm⁻³ or less. Further, it is recommended that the concentration of a mobile ion such as sodium, lithium, or potassium contained in a semiconductor film be 5×10¹⁸ cm⁻³ or less.

[00031] In order to achieve the above mentioned objects, it is desired to use a gas having a high purity of 99.999% or more, paying a full attention to a material gas. Further, according to the present invention, the insulating substrate containing a lot of mobile ion sources is born in mind to be used. Then, in order to conduct the present invention more efficiently, it is preferable to cover without omission all the circumference of insulating substrate, when the first silicon nitride film is formed on such substrate. In such situation, it is possible to exceedingly lower a probability that a mobile ion mixes into an device region from the substrate.

[00032] FIG. 4 shows an instance of forming a lightly doped drain (LDD), which is a well known prior art, according to the present invention. First of all, a silicon nitride film 402 is formed on such insulating substrate 401 as quartz or AN glass in 50 to 1000 nm, using a low pressure CVD method. At this time, by covering all the surface of substrate including its back side too, as aforesaid, the probability that a mobile ion generated in the back side reaches to the surface at a later process, for example, the probability that a mobile ion intrudes into a film during a gate oxide film forming, will be remarkably lowered. Also, it is preferable in keeping the manufacturing equipments purified. A silicon oxide film 403 used in buffering is formed on a silicon nitride film in 50 to 1000 nm thick, also using a low pressure CVD method. At this time, a halogen element such as chlorine, fluorine will be involved in the obtained silicon oxide film, if a gas containing such halogen as hydrogen chloride (HCl), nitrogen fluoride (NF₃ or N₂F₄), chlorine (Cl₂), fluorine (F₂), all sorts of fleon gas, and carbon tetrachloride (CCl₄) is mixed in a raw material gas, in a ratio of 3% to 6% by volume, e.g. 5% by volume.

[00033] This halogen combines with such alkaline ion as sodium, to fix sodium, then it is very effective to prevent sodium contamination. But it is not preferable that an excess addition of a halogen makes a film rough and damages its adhesion property and surface flatness. Also, it is recommendable to mix the above gas containing a halogen element in a raw material gas in a ratio of 2 to 5% by volume, in case of forming the said film using a light CVD method or a plasma CVD method, instead of a low pressure CVD method. Further, it is recommendable to mix the above halogen gas in a sputtering atmosphere by 2 to 20% by volume, in case of forming the said film by a sputtering method. Since the composition in an atmosphere does not be reflected to a film composition, it is needed to increase a little more concentration of halogen gas than the case of CVD methods.

[00034] Then, an amorphous silicon film, a microcrystal or a polycrystal silicon film is formed in contact with the silicon oxide film 403, in 20 to 500 nm thick, by means of a low pressure CVD method, a plasma CVD method, or a sputtering method. Then, this is etched to form the island shape. When the silicon film is formed, it is preferable to introduce a halogen element into the film, in the same way as the forming of the aforesaid film 403. As a method to introduce a halogen element, it may be possible to mix a gas containing a halogen into a film forming atmosphere, in the same method as the aforementioned film 403, or to introduce the gas after the film forming, by an ion implantation method. At this time, it is required to control a raw material gas concentration so that a halogen element concentration in a film be in the range of 1×10¹⁸ to 5×10²⁰ atoms/cm³, preferably 1×10¹⁹ to 1×10²⁰ atoms/cm³.

[00035] Still further, at the same time, if a hydrogen element concentration in a film is four times as much as or less, preferably one time as much as or less than that of a halogen, a halogen adding effect will be raised. This effect can be explained as follows: Hydrogen is necessary to terminate a dangling bond of silicon, but its combination is weak and is simply loosed. On the other hand, halogen element combines with silicon strongly. If hydrogen exists in a silicon excessively, which is meant by that there are a lot of dangling bonds in a film, almost all the halogen will combine with silicon. Then, it is impossible to effect gettering of a mobile ion traveling in a film. Therefore, it can be supposed that in a silicon containing much hydrogen, the halogen adding effect will be lowered. To the contrary, in a silicon containing less hydrogen, the halogen adding effect will be raised vice versa.

[00036] Also, it is desired that as a harmful element except a mobile ion, the concentration of carbon, nitrogen, and oxygen be 7×10¹⁹ atoms/cm³ or less, preferably 1×10¹⁹ atoms/cm³ or less, in such semiconductor film as silicon. This is resulted from the reason that these elements can not be removed by adding halogen.

[00037] Further, though the gettering of such mobile ion as sodium lithium, and, potassium can be effected by adding halogen, this effect will be denied in case where these elements exist in excess. Therefore, it is preferable that the concentration of each mobile ion be 5×10¹⁸ atoms/cm³ or less.

[00038] Then, on a silicon film formed as above mentioned, a silicon oxide film as a gate insulating film is formed, in 10 to 200 nm, preferably 10 to 100 nm, using a low pressure CVD method or a sputtering method. In this case, it is also preferable to mix a halogen containing gas beforehand in a raw material gas or a sputtering gas, in the same way as mentioned before.

[00039] Next, on this silicon oxide film, a polycrystal or a microcrystal silicon film, in which about 10²¹ cm⁻³ of phosphorus is doped, is formed using a low pressure CVD method or a plasma CVD method. Then, the silicon film and the gate insulating film (silicon oxide) provided beneath the silicon film are patterned to form a gate electrode 410 and a gate insulating film 404.

[00040] Further, a source (drain) region 407 and a drain (source) region 409, which contain comparatively low impurity concentration (about 10¹⁷ to 10¹⁹ cm⁻³), are formed, implanting an ion in a self-alignment method, using this gate electrode as a mask. The portion where an impurity is not implanted remains as a channel region 408. Thus, FIG. 4(A) is obtained.

[00041] Next, as shown in FIG. 4(B), a PSG (phosphosilicate glass) film 413 is formed throughout the surface, using a low pressure CVD method. And this is etched by the conventional direction depending etching method to form a side wall 414, in the side of a gate electrode. Then, a source (drain) region 407a and a drain (source) region 409a, which contain a high impurity concentration, are formed by implanting an ion again. A region having a low impurity concentration becomes a source (drain) region 407b and a drain (source) region 409b, and forms LDD (lightly doped drain). Thus, FIG. 4(C) is obtained.

[00042] After that, as shown in FIG. 4(D), a silicon nitride film 405 is formed all over the surface, in 50 to 1000 nm thick using a low pressure CVD method. Then, a silicon film is crystallized to activate a source, drain region, by a low temperature annealing of e.g. 600° C. This process can be effected by a laser annealing. Thus, TFT intermediate is obtained.

[00043] FIG. 4 shows one preferred example of the present invention, and the present invention is not limited to the above process. In the example of FIG. 4, a silicon nitride film, a gate electrode, and a source or a drain region do not adjoin, like the example of FIG. 3. That is, the case of FIG. 4 is different from that of FIG. 2 in that a side wall 414 is provided. Then, there is no problem which was worried about in FIG. 2. Further, the case of FIG. 4 is characterized in that an adding of a donor or an acceptor can be easily effected without passing an insulating film, in different way from the case of FIG. 3.

[00053] The present invention will be explained in more detail by reference to the following Examples, in connection with the drawings. The present invention is by no means limited to the Examples.

[00054] The characteristics of TFT in accordance with the present invention is described as follows. TFT used in this Example was a LDD type TFT prepared on a quartz glass, according to the process of FIG. 4. First of all, a silicon nitride film 402 was formed in 100 nm thick, on a quartz glass 401 and also a back and a side thereof (i.e. in whole surface of the substrate), using a low pressure CVD method. Then, continuously by this low pressure CVD method, a silicon oxide film (insulating film) 403 (a low temperature oxidation film, LTO film) was formed in 200 nm thick. Lastly, by this low pressure CVD method as well, an amorphous silicon film was formed on the silicon oxide film 403 in 30 nm thick. At this time, the maximum process temperature was 600° C.

[00055] Next, the amorphous silicon was patterned into the island shape. Then, the very thin surface portion of the amorphous silicon was oxidized, in 2 to 10 nm thick by an anodic oxidation method. The oxidized surface portion comprised silicon oxide. Then, a silicon oxide film (insulating film) was formed thereon in 100 nm by a sputtering method. Here, the sputtering atmosphere was mixed gases consisting of oxygen and argon, or other rare gas, and the oxygen partial pressure thereof was 80% or more. At this time, a defect generated in a surface film by a shock of sputtering. For example, if the surface film is a silicon film, oxygen atom is implanted in the silicon to increase an oxygen concentration. Under this status, silicon will be in a high localized level. Namely, a boundary between silicon and silicon oxide will be unclear. However, according to the present Example, the beforehand forming of the thin anodic oxidized film can prevent the above atom mixing, and then the boundary between silicon and silicon oxide can be kept clear. This is due to the reason that there exists silicon oxide in advance in sputtering.

[00056] After forming of this silicon oxide film, an n⁺-type microcrystal silicon film containing 10²¹ cm⁻³ degree of phosphorus was formed in 300 nm thick, using a low pressure CVD method. The maximum process temperature during the above film forming was 650° C. Then, patterning was carried out to form a gate electrode 410 and a gate insulating film 404. The gate electrode 410 was made from the n⁺-type microcrystal silicon film and the gate insulating film 404 was made from the anodic oxidized surface portion and the silicon oxide film formed thereon. Further, a source and drain regions 407 and 409 were formed, by implanting 2×10¹⁸ cm⁻³ of arsenic ion using an ion implantation. Thus, FIG. 4(A) was obtained.

[00057] As shown in FIG. 4(B), PSG (phosphosilicate glass) film 413 was formed by a low pressure CVD method, and as shown in FIG. 4(C), a side wall 414 was formed by a direction depending etching method. Then, 5×10²⁰ cm⁻³ of an arsenic ion was implanted in the regions 407a and 409a, by an ion implantation.

[00058] After that, a silicon nitride film 405 was formed all over the surfaces by a low pressure CVD method. Thereby, FIG. 4(D) was obtained. Then, it was annealed at 620° C. for 48 hrs. in a vacuum to activate the regions 407a, 407b, 408, 409a, and 409b. Then, as an interlayer insulation, PSG (phosphosilicate glass) film was formed all over the surface by a low pressure CVD method. And a hole was made to provide an aluminum electrodes in the source and the drain regions. Lastly, for the purpose of a passivation, a silicon nitride film was again formed all over the surface, using a low pressure CVD method.

[00059] TFT prepared in this way as mentioned above was remarkably high in confidence. It showed that the operation features of an device did not change by even a so-called Bias-Temperature treatment. (BT treatment). The features are illustrated in FIG. 8. BT treatment was conducted by applying a bias voltage V_B between the gate (G), the source(S), and the drain(D) of a circuit shown in FIG. 8 under heating. Concretely, as soon as TFT was prepared, the gate voltage−drain current characteristic of the TFT was measured at room temperature (V_B=0). Then, +20 V of voltage was applied to the gate electrode at 150° C. for an hour and the measurement of the gate voltage−drain current characteristic of the TFT at room temperature was conducted (V_B=+20 V). Next, −20 V of voltage was applied to a gate electrode again at 150° C. for an hour, then the gate voltage−drain current characteristic of the TFT was measured at room temperature (V_B=−20 V), and the fluctuations of threshold value voltage of TFT was examined. FIG. 8(B) shows a TFT features obtained by the method described in the above. Like this, the features were not dependent on the bias voltage V_B, on TFT according to the present Example, and the fluctuations of threshold value voltage was 0.2 V or less by the accurate measurement.

[00060] On the other hand, FIG. 8(A) shows TFT obtained by the same process with the present Example, excepting that the silicon nitride films 402 and 405 were not provided. As evident in FIG. 8(A), TFT features are very dependent on the V_B. Such fluctuations (fluctuations of threshold value) are explained to be caused by such mobile ion as sodium etc. in a gate insulating film. Namely, it is explained that the more mobile ion is contained, the more fluctuations are effected, and that the case where the fluctuations are few as shown in FIG. 8(B) are due to a small amount of the mobile ion. It is estimated that a mobile ion amount in a gate electrode of TFT prepared by the present Examples is about 8×10¹⁰ cm⁻³, from the fluctuations width of threshold value voltage. Thus, it is possible to improve the features of TFT remarkably and to raise its confidence by providing a silicon nitride film in accordance with the present invention. cl EXAMPLE 2

[00061] TFT used in this Example was a LDD (lightly doped drain) type TFT, which was prepared on a quartz glass substrate according to the process of FIG. 4. Firstly, a silicon nitride film 402 was formed on a quartz glass substrate 401 and a back thereof, in 100 nm thick by a low pressure CVD method. Then, a silicon oxide film (low temperature oxidation film, LTO film) 403 was formed continuously in 200 nm thick by a low pressure CVD method. Lastly, an amorphous silicon film was formed thereon, in 30 nm thick by also a low pressure CVD method. At this time, the maximum temperature of the process was 600° C. In this process, the film forming was carried out in a CVD apparatus consisting of three reactors arranged in succession, but in case of the forming of a silicon oxide film and an amorphous silicon film, a reaction was effected by adding 5% by volume of hydrogen chloride gas (HCl) as a halogen additive gas, in addition to a raw material gas.

[00062] As a result, a halogen (chlorine) could be added in a silicon oxide film and an amorphous silicon film. By a secondary ion mass spectrometry, the chlorine contents in a silicon oxide film and an amorphous silicon were 2.3×10¹⁹ atoms/cm³, 3.1×10¹⁹ atoms/cm³, respectively. As a raw material gas of a silicon nitride film, dichlorosilane (SiCl₂H₂) and ammonia (NH₃), as a raw material gas of a silicon oxide film, disilane (Si₂H₆) and oxygen (O₂) and hydrogen chloride, and as a raw material gas of an amorphous silicon, disilane and hydrogen chloride were employed respectively. The purity of each raw material was 6N The contents of hydrogen atom in a silicon oxide film and an amorphous silicon film thus obtained were confirmed to be both 1×10¹⁹ atoms/cm³ or less. Also, since the film forming was effected continuously without being exposed in the air, the concentration of carbon, nitrogen, and oxygen in the silicon film was confirmed to be 1×10 atoms/cm³ or less.

[00063] Next, an amorphous silicon film was patterned into an island shape, and the very thin surface portion of the amorphous silicon was oxidized, in 2 to 10 nm thick by an anodic oxidation. The anodic oxidation was effected, using a platinum electrode as a cathode in an electrolyte containing N-methylacetoamide (NMA) or tetrahydrofurfurylalcohol (THF) added with KNO₂, by a constant voltage method at 10 to 50° C. After completion of the anodic oxidation, it was annealed at 600° C. in an argon atmosphere. Then, a silicon oxide film was formed thereon in 100 nm thick using a sputtering method. Here, as the sputtering atmosphere, a mixed gas of oxygen and argon or other rare gas and hydrogen chloride, in which the oxygen partial pressure was 80% or more and the hydrogen chloride concentration was 10%, was employed.

[00064] In the film forming by sputtering, a defect generated in a surface film by a shock of sputtering. For example, if the surface film is a silicon film, oxygen atom is implanted in silicon to increase oxygen concentration therein. Under such situation, silicon will be in a lot of the localized level. That is, a boundary between silicon and silicon oxide will be unclear. However, according to the present Example, by beforehand forming of the thin anodic oxidized film, the above mentioned atom mixing can be prevented, and the boundary between the silicon film and the silicon oxide film can be maintained, because the silicon oxide film (the thin anodic oxidized film) has been formed before the sputtering.

[00065] After forming of this silicon oxide film, an n⁺-type microcrystal silicon film containing 10²¹ cm⁻³ extent of phosphorus was formed, in 300 nm thick using a low pressure CVD method. The maximum process temperature in the above film forming was 650° C. Then, a gate electrode was patterned to form a gate electrode 410 and a gate insulating film 404. Further, a source and a drain regions 407 and 409 were formed, implanting 2×10¹⁸ cm⁻³ of arsenic ion by ion implantation method. Then, FIG. 4(A) was obtained.

[00066] Next, as shown in FIG. 4(B), PSG (phosphosilicate glass) film 413 was formed by a low pressure CVD method, and a side wall 414 shown in FIG. 4(C) was formed by a direction dependent etching. Further, 5×10²⁰ cm³ of arsenic ion was implanted in the regions 407a and 409a by ion implantation method.

[00067] After that, a silicon nitride film 405 was formed throughout the surface, by a low pressure CVD method. Thereby, FIG. 4(D) was obtained. Then, it was annealed to activate the regions 407a, 407b, 408, 409a, and 409b, at 620° C. for 48 hrs. in a vacuum. As an interlayer insulation, PSG (phosphosilicate glass) film was formed all over the surface by a low pressure CVD method, and a hole for an electrode was made to form an aluminum electrodes in a source region and a drain region. Lastly, for the purpose of a passivation, a silicon nitride film was formed again throughout the surface, by a low pressure CVD method.

[00068] TFT prepared in this way, was remarkably high in a confidence. It pointed out that the features of device operation did not change by a so-called bias-temperature treatment (BT treatment). BT treatment is meant by the treatment that a voltage is applied between a source and a drain and to a gate electrode in heating. There is no problem in case of a normal device, but the features change is seen in case of a device in which for example, a mobile ion is contained. Such situation is shown in FIG. 6.

[00069] FIG. 6(A) indicates TFT in which a mobile ion exists in a gate insulating film and in a channel region. In the channel region, there exists an alkaline mobile ion (indicated as A⁺ in FIG.) which functions as a donor. Then, the channel shows a weak N type (N⁻-type), the state of which is referred to as state 1. By applying positive bias voltage to the gate electrode and between the source and drain of this TFT, as shown in FIG. 6(B), the mobile ion (positive ion) in the channel region will be made far away from a gate electrode in the first place, and then, the channel region will be made intrinsic (I-type). This state is referred to as state 2. Consequently, I_D (drain current)−V_G (gate voltage) characteristic of TFT will be moved largely to the rightside, as shown in FIG. 6(D).

[00070] However, in case where a mobile ion exists in a gate insulating film, the mobile ion gathers to the under part of a gate electrode (to the side of a channel region), owing to a bias voltage applied to a gate electrode. As a result, the channel region will have a positive electric field, and electron will gather therein. Then, the channel region shows a weak N type again. This state is referred to as state 3. As shown in FIG. 6(E), I_D−V_G curve shifts to the left side from the state 2 to the state 3. After all, the curve has been shifted to the right side, compared with the first one. Also, in case where a negative bias is applied conversely, a mobile ion gathers in a channel region. Then, the channel region will proceed to N type, resulting in the state that the drain current can not be controlled by the gate voltage.

[00071] Concretely in the present Example, the gate voltage—drain current characteristic of TFT at room temperature was measured (V_B=0), immediately after TFT was prepared. Then, +20 V of voltage was applied to a gate electrode at 150° C. for an hour, and the gate voltage—drain current characteristic of TFT was measured at room temperature (V_B=+20 V). After that, −20 V of voltage was applied again to a gate electrode at 150° C. for an hour. Then, the gate voltage−drain current characteristic of TFT was measured at room temperature (V_B=−20 V), and the threshold value voltage fluctuation of TFT was examined.

[00072] The characteristic of TFT prepared by the method as described in the above is illustrated in FIG. 7(B). The bias voltage has no influence on TFT at all as such, and the fluctuation of threshold value voltage was 0.2 V or less as a result by an accurate measurement.

[00073] On the other hand, TFT shown in FIG. 7(A) is the one in case where the silicon nitride films 402 and 405 are not provided, and the halogen concentration of each film thereof is 1×10¹⁴ cm⁻³ or less, and also it is prepared by the same process with the method disclosed in the present Example, excepting the said points. As evident in the FIG. 7(A), the characteristic is dependent upon V_B to a large extent. It is estimated that the content of a mobile ion in a gate electrode of TFT prepared by the present Example is in a degree of 8×10¹⁰ cm⁻³, from a fluctuation width point view of the threshold value voltage of FIG. 7(B).

[00074] After the above measurement, the concentrations of sodium, potassium, and lithium in a silicon film (channel region) and a gate insulating film of TFT prepared by the present Example were measured to be 3×10¹⁷ cm⁻³, 7×10¹⁵ cm⁻³, and 5×10¹⁵ cm⁻³ respectively. Though such a lot of alkaline element existed therein, the mobile ion amount was little. This fact is presumed to be caused by that it was fixed with halogen (with chlorine in this case). The measurement of sodium, potassium, and lithium concentration in TFT prepared for a comparison showed such much amount as 7×10¹⁸ cm⁻³, 2×10¹⁶ cm⁻³, and 4×10¹⁹ cm⁻³ respectively.

[00075] Thus, it can be presumed that the blocking effect of a silicon nitride film in accordance with the present invention is obtained as stated above. Namely, it is understood that the features and the confidence of TFT can be improved remarkably.

[00076] By the present invention, it is possible to prepare a thin film semiconductor device such as TFT, on which the influence of such a mobile ion as sodium can be reduced. And it is achieved that TFT can be prepared, even on a substrate, the device on which has not been able to be prepared so far, on account of the existence of a mobile ion therein. The present invention can be applied to a coplanar type as shown in FIGS. 1 to 4, an inverse coplanar type, a stagger type, and an inverse stagger type. Also, since the present invention does not impose restrictions on an operation of a thin film semiconductor device, the silicon of a transistor may be an amorphous, a polycrystal, a microcrystal, a single crystal and an intermediate state thereof.